Abstract
Keywords
Introduction
Inadequate sexual and reproductive health (SRH) and heightened gender-based violence (GBV) pose considerable threats to public health and human rights (Glasier et al., 2006). The WHO’s broad definition of SRH encompasses GBV: SRH is a state of physical, emotional, mental, and social wellbeing in relation to sexuality and the reproductive system, including the ability to have pleasurable and safe sexual experiences, free of discrimination, coercion, or violence (WHO, n.d.). GBV encompasses a variety of damaging acts perpetrated against someone based on their gender expression, gender identity, or perceived gender (IASC, 2015). Thus, SRH and GBV are interrelated: experiencing GBV has been associated with unintended pregnancy, complications during pregnancy, pain during sexual intercourse, reproductive tract infections, and sexually transmitted infections (STIs) (Jina & Thomas, 2013; Meinhart et al., 2021). Moreover, inadequate SRH services and education can magnify GBV harms and increase the risk of GBV perpetration (Igras et al., 2014; Kågesten et al., 2016).
Health disparities leading to differential SRH and GBV outcomes are influenced by gender inequality and thus disproportionately affect women and girls (Darmstadt et al., 2019). Moreover, risk factors for adverse SRH and GBV are heightened in humanitarian and fragile settings (IASC, 2019). Humanitarian and fragile settings are often associated with periods of chronic stress, poverty, conflict, forced displacement, strained social support networks, as well as the loss of medical and public health infrastructure which erode the provision of SRH care and exacerbate GBV (Classen et al., 2005; Logie et al., 2019). The COVID-19 pandemic has further increased women and girls’ vulnerability to GBV and has led to disinvestment in lifesaving SRH services (Asi et al., 2022; Carter et al., 2020; Stark et al., 2020). This pattern of magnified GBV risk and reduced SRH service availability is particularly pronounced in humanitarian and fragile settings (Lokot & Avakyan, 2020; Tran et al., 2020).
Inequitable vaccine distribution to humanitarian and fragile settings heightens community transmission, contributes to COVID-19-related population morbidity and mortality, and accelerates the overcapacity of health and social services, all of which can yield negative consequences on GBV and SRH (Lobkowicz et al., 2021; Zard et al., 2021). Further, emerging literature from high income settings illustrates that the COVID-19 pandemic has magnified reported gender-based violence (Piquero et al., 2021) and that survivors of domestic violence are at an increased risk of COVID-19 diagnosis (Chandan et al., 2021). In low-income countries, the pandemic has negatively affected social determinants of health such as household income, unemployment, and food and food insecurity, known risk factors for GBV and adverse SRH (Bourgault et al., 2021). In humanitarian and fragile settings, school closures and lack of socioeconomic supports are particularly threatening for adolescent girls who face increased risk of sexual exploitation, transactional sex, and child marriage, all of which can negatively affect SRH through unintended pregnancy, STI, or gynecological trauma and may have life course and intergenerational health consequences (UNFPA, 2020; Yukich et al., 2021).
Researchers are increasingly interested in drawing on primary data collection in order to address the unintended gendered consequences of the COVID-19 pandemic in humanitarian and fragile settings. However, data collection in such settings is difficult in the best of times due to the instability inherent within natural disasters, low literacy, conflict, urban poverty, mass displacement, and civil unrest (Bennouna et al., 2017; Stark & Ager, 2011a). To further compound data collection difficulties, the pandemic context presents considerable methodological and ethical concerns, whereby risks may outweigh the potential benefits of primary data collection (WHO, 2020). For example, in-person primary data collection may increase the risk of community transmission considering the lack of COVID-19 vaccine equity in humanitarian and fragile settings (Y. Liu et al., 2020; Singh et al., 2020). In addition, stay-at-home orders increase the proximity of family members, thereby increasing the risk that research-related disclosures of sensitive information pertaining to GBV or SRH will be heard and that violence or stigmatization may ensue (Peterman et al., 2020; Seff et al., 2021).
A growing number of researchers are turning to digital and remote technologies to collect data on GBV and SRH during the pandemic. Promising remote data collection (RDC) tools include instant messaging platforms, phone interviews, online surveys, and mobile applications (Emezue, 2020; Hensen et al., 2021). While the deployment of such tools rapidly expanded during the COVID-19 pandemic, these modes of data collection pose their own methodological and ethical challenges (Seff et al., 2021). There is a considerable lack of research synthesis on the methodological and ethical consequences of RDC and this is especially true for RDC on SRH and GBV-related topics. Understanding best practices and pitfalls of RDC is critical to safeguarding the SRH rights and safety of women and girls. This systematic review is guided by the following research question: What is the current peer-reviewed evidence on the use of remote data collection tools in humanitarian and fragile settings to collect primary data on women and girls’ knowledge, attitudes, behaviors, and experiences related to SRH and GBV? Specifically, within the context of humanitarian and fragile settings this systematic review seeks to identify: (i) modes of RDC used to capture SRH and GBV and (ii) primary concerns pertaining to upholding methodological rigor during data collection/measurement and ethical standards.
Methods
We conducted a mixed methods systematic review in March 2021 to identify qualitative, quantitative, and mixed methods studies conducted in humanitarian and fragile settings that collected data on women and girls’ SRH and GBV using RDC tools. The methodology is guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses standards (Page et al., 2021).
Eligibility Criteria
Inclusion/Exclusion Criteria.
Similar to other systematic reviews, humanitarian and fragile settings were defined as low- or middle-income countries involving any of the following: refugees, internal displacement, armed conflict, civil unrest, natural disasters, and poverty (Rubenstein et al., 2020; Stark et al., 2021; Stark & Ager, 2011b; Vu et al., 2014). Our operationalization of humanitarian and fragile settings was informed by definitions of (i) humanitarian response to crises: “range of situations including natural disasters, conflict, slow and rapid onset events, rural and urban environments, and complex political emergencies” (The Sphere Project, 2011, p. 9); (ii) complex emergencies where considerable breakdown of authority resulting from internal or external conflict requires an international response beyond the or capacity of a single agency (ELRHA, n.d.); and (iii) state fragility wherein governments cannot or will not deliver core functions of poverty reduction, safety and security, and capacity to manage public resources and basic services (Cammack et al., 2006).
Identified studies that met the definitions of humanitarian response, complex emergency, or fragile setting based on the setting description and the team’s collective expertise were then cross-referenced using annual reports of humanitarian-related funding from Office for the Coordination of Humanitarian Affairs (OCHA), reports detailing influx of refugees from United Nations High Commissioner for Refugees (UNHCR), the presence of peacekeeping operations United Nations Security Council’s Department for Peacekeeping, and the World Bank list of fragile states. RDC was operationalized as the use of technology to facilitate primary data collection in a way that replaces non-physically distanced in-person interaction between participants and enumerators (Hensen et al., 2021; Seff et al., 2021). Only studies that employed RDC tools across all stages of data collection were included because we were interested in synthesizing studies that primarily employed RDC tools and were designed a-priori to collect data remotely.
The outcome of interest was women and girls’ SRH, inclusive of GBV. We used the WHO’s broad definition of SRH that encompasses GBV, wherein SRH is a state of physical, emotional, mental and social wellbeing in relation to sexuality and the reproductive system, including the ability to have pleasurable and safe sex experiences, free of discrimination, coercion, or violence (WHO, n.d.). GBV was conceptualized as violence perpetrated against someone based on their gender expression or identity, such as intimate partner violence (IPV), non-partner sexual violence, child marriage, sexual abuse and exploitation, and reproductive coercion, female genital cutting. Any outcome related to sexual health, reproductive health, and/or GBV was included so long as data were collected from women and girls about their own experiences. Studies that included mixed-gender samples were also considered.
Search Protocol and Study Selection
The published literature was searched using strategies created by a medical librarian. The search strategies were established using a combination of standardized terms and keywords. To identify humanitarian and fragile settings, keywords were combined with a list of countries from the Uppsala Conflict Data Program’s publicly available datasets on settings experiencing armed conflict from 1989 to 2019 (UCDP, n.d.) and the International Refugee Commission’s 2020 and 2021 emergency watchlist (International Rescue Commitee, 2020; International Rescue Committee, 2019). The search was run in March 2021 without filters or limits in the databases Ovid Medline 1946-, Embase 1947-, Web of Science 1900-, Clinicaltrials.gov, and Scopus 1960. Sources were exported to Endnote and duplicates were removed. Full electronic search strategies are provided in Appendix A.
The screening was conducted using Covidence (Covidence, 2021) and occurred in two stages: (i) title and abstract and (ii) full text review. A team of reviewers worked independently to review titles/abstracts (one reviewer per title/abstract) and the same team worked in duplicate to review full texts (two reviewers per full text). During the title/abstract phase, team consensus was sought if a team member was unsure about how to apply eligibility criteria. Sources where titles/abstracts met the inclusion criteria underwent full text review. During the full text screening, discordance between the reviewers was adjudicated by a third, independent reviewer.
Quality Assessment and Data Extraction
Data extraction and quality assessment were performed in duplicate using Google spreadsheets by the team of reviewers and disagreements were adjudicated by consensus. Appendix B contains the data extraction and quality assessment template in table form. Included studies were extracted and assessed by study design according to three main themes: general characteristics, methodological rigor, and ethics. Sources were first cataloged according to general information: authors, year of publication, country, purpose, study design, sample size, sample characteristics, participants’ minimum age, RDC instrument employed, whether the study was conducted during the COVID-19 pandemic, and outcome measures.
Methodological quality and rigor were assessed by study design (qualitative, randomized controlled trials, quantitative non-randomized, quantitative descriptive, and mixed methods), using selected criteria from the Mixed Methods Assessment Tool (MMAT) that related specifically to data collection, measurement, and sampling considerations (Hong et al., 2018; Pace et al., 2012). We also further examined methodological rigor by extracting information on sampling and recruitment strategies, risk of selection bias for quantitative studies, discussion of saturation (qualitative studies), and noted limitations/challenges with RDC. These additional fields were incorporated to better address our research question of identifying primary concerns pertaining to data collection and measurement using RDC tools. External sources outlining best practices on RDC of sensitive information were consulted to develop data extraction fields pertaining to ethics (Bhatia et al., 2020; National Network to End Domestic Violence, n.d.; Peterman et al., 2020; Seff et al., 2021). Sources were reviewed based on procedures to obtain informed consent, inclusion of referral services, incorporation of measures to safeguard participant safety, discussion of adverse events, data security and confidentiality, participant literacy concerns, and the gender digital divide.
Results
Study Selection
A total of 5,772 citations were identified with the search strategy and exported to Endnote. 2857 duplicates were removed, resulting in a total of 2915 unique citations. The 2915 sources were screened in the title and abstract phase. In total, 424 sources were assessed for full-text eligibility based on the inclusion and exclusion criteria and 21 were included for extraction (refer to Figure 1 for the PRISMA flow diagram). PRISMA flowchart of study selection process.
Study Characteristics
Characteristics of Included Studies by Study Design.
* Conducted during the COVID-19 pandemic.
Note. RDC = remote data collection; N = sample size.
Qualitative Synthesis of Results
Methodological Rigor by Study Design.
* RDC = remote data collection; ** Items from the MMAT tool; *** RCT protocol.
Ethical Considerations by Study Design.
* RDC = remote data collection.
Methodological Rigor: Quantitative Descriptive
Non-probability sampling using online surveys comprised the predominant form of sampling among the prevalence studies (Abuhammad, 2021b; Aolymat, 2021; Baloushah et al., 2019; Fakunmoju & Bammeke, 2013; Ghimire et al., 2020; González-Hernández et al., 2020; Maasoumi et al., 2019; Mahmood et al., 2021; Restar et al., 2020). However, given that most quantitative descriptive studies sought to estimate prevalence of GBV or SRH outcomes or assessed factors associated with GBV and SRH at a population level, the rigor of non-probability sampling is limited by the threat of overt and hidden selection biases which reduce confidence in prevalence estimates and the measures of association obtained. Online recruitment posed concerns for response rates, selection bias, and sample representation. Five of the nine studies that employed online surveying did not report response rates (Aolymat, 2021; Baloushah et al., 2019; Maasoumi et al., 2019; Mahmood et al., 2021; Restar et al., 2020). Even when reported, only one study made the distinction between how many potential participants received the study invitation, number of potential participants who engaged with the online survey link, the final number of participants who completed the survey, and the number of participants who were included in the study sample (González-Hernández et al., 2020). The lack of a well-defined sampling frame is the main methodological challenge with online surveys disseminated though social media/networking websites and SMS platforms. Consequently, it is unclear to what extent estimates are generalizable to the population of interest or are affected by overestimation, underestimation, or attenuation. Moreover, many studies noted that online surveys systematically exclude women and girls without internet access, who are not active on social media, or do not use messaging platforms.
Other studies recruited participants through existing contraception providers via purposeful sampling (Lebetkin et al., 2014), leveraged participants’ social networks though respondent driven sampling (Memiah et al., 2020), employed purposeful sampling to select participants who experienced childhood sexual abuse (criteria of interest) and accessed a child advocacy center (Koçtürk et al., 2019) and included all available users of the mobile for Reproductive Health (m4RH) program within the sample (L’Engle et al., 2013). We deemed sampling as being affected by bias when it was solely contingent on access to a RDC tool and excluded users who could not be reached or did not respond to survey questions after providing consent. Systematic differences between participants who self-select out of the study due to no response or inability to be contacted pose risks for selection bias. In terms of measurement, four studies were deemed to be affected by measurement bias given these studies did not: use validated survey instruments, report measures of reliability/validity (Ghimire et al., 2020; Mahmood et al., 2021), pilot test survey instruments (Koçtürk et al., 2019), or detail how interview guides were developed (Lebetkin et al., 2014). The use of SMS also limited the number and depth of survey questions posted to participants (L’Engle et al., 2013).
Methodological Rigor: Mixed Methods
Two of the three mixed methods studies (Ko-Ko-Zaw et al., 2011; Vahdat et al., 2013) employed availability sampling; this was deemed sufficient given the primary goal was to assess the feasibility, use, and perceptions of SRH services/resources among existing users. However, even among users and hotline, adequate response rates were not guaranteed: some participants were users of the service/resource but did not respond to the survey. Ybarra et al. (2020) employed mixed methods (FGDs, open text responses, Likert responses) to develop and pilot an HIV prevention program for Ugandan youth and recruited participants through social media advertisements. To increase the likelihood of obtaining a diverse sample by gender, education, income, and sexual experience, interested participants were first screened for demographic characteristics.
We noted declining participant retention between the quantitative and qualitative components of data collection in Vahdat et al.’s (2013) study. The authors noted that delays between user interaction with the SRH service and participation in research reduced retention. Qualitative data capture was also limited to simple description, as opposed to in-depth probing and contextualization; quantitative data was limited to basic demographic questions and questions on perceptions/experiences that could be collected via SMS (Vahdat et al., 2013). Ybarra et al. (2020) noted positive aspects of online bulletin board facilitated FGDs; however, there was no discussion of probing participants for more information. Moreover, multiple phone calls/SMS were needed to encourage participation and resolve technology issues, some phones were unable to open the survey, and SMS was preferred over online surveys due to higher internet costs. Overall, the measures used were not validated or relied on service use data (Ko-Ko-Zaw et al., 2011; Vahdat et al., 2013; Ybarra et al., 2020).
Methodological Rigor: Quantitative Non-Randomized Evaluations
Two quantitative non-randomized evaluations were identified. The only study to collect a probability sample was conducted by Doubova et al. (2017). A simple random sample of public secondary schools from a defined geographical location in the Iztapalapa Delegation of Mexico was obtained, and students were selected from within the respective schools. The authors tried to quantify and correct for the potential effect of social desirability bias due to self-report by employing the “Lie Scale” of the Eysenck Personality Questionnaire, which measures attempts to respond in socially desirable ways using 20 dichotomous questions (Eysenck & Lara-Cantú, 1992). The second quantitative non-randomized evaluation utilized purposeful and criterion sampling to recruit chemical contraceptive (DMPA-SC) providers in Nigeria who then identified potential participants (J. J. Liu et al., 2018). The authors noted that the quality measure employed may not have been suitable for the Nigerian context and opted for binary response choices for telephonic interview due to low quality phone connection and research fatigue.
Methodological Rigor: RCT
For the RCT protocol (O. L. McCarthy, Osorio et al., 2017) and later completed RCT (O. O. McCarthy et al., 2018) and cluster RCT (Babalola et al., 2019), eligibility was contingent on having access to or owning a phone. McCarthy et al. (2017) further limited eligibility to Android phones. Given that RCTs collect data at baseline and follow-up periods, different RDC tools can be used: researchers offered a choice between online/phone for baseline and mobile application for end line (2017, 2018). The Babalola et al. (2019) RCT was negatively affected by high and unbalanced attrition rates. This was attributed to the research team not using an attention-matched control group; there were a greater number of calls received by the intervention group compared to the control group. Babalola et al. (2019) also noted that survey calls sounded like telemarketing calls and technical issues further demotivated participation.
Methodological Rigor: Qualitative
The single qualitative study identified recruited participants from the first author’s WhatsApp list using maximum variation sampling (Mbulayi et al., 2021). WhatsApp was also used to facilitate semi-structured interviews using the voice notes feature with question prompts sent over text/voice notes. However, given the authors did not detail the limitations of WhatsApp-based interview facilitation, it is not clear how this WhatsApp-based qualitative interviewing method may affect data quality.
Ethics
Ethics were assessed both in terms of procedural ethics (mandated requirements such as informed consent and referral services) and issues concerning the equity and inclusion of marginalized groups in research. Across the design types, the process of obtaining informed consent was enacted through the type of RDC tool employed. Two studies employed additional mechanisms through which to collect informed consent and assess capacity to consent including, a secure online database (O. McCarthy et al., 2017) and an online cognitive screening form to assess comprehension (Restar et al., 2020). Two studies collected informed consent in-person even though data collection occurred remotely (Babalola et al., 2019; Doubova et al., 2017). Generally, the sources did not report how participants could ask questions during the consent process when relying on SMS, online surveys, or mobile applications to collect data (although this could be reflective of space limitations rather than gaps in the consent protocols). However, Ybarra et al. (2020) did contact interested participants by phone to further explain the research and obtain informed consent, thereby offering an opportunity to answer questions. Also, no sources reported on how information on data security and privacy were communicated to participants, particularly when personal information (geo-location and phone numbers) were collected and given that data were automatically received in survey platforms via processes that may be foreign to participants. In terms of data confidentiality/security, studies employed password protected data storage. When personally identifiable information was collected, separate databases storage was utilized (O. McCarthy et al., 2017).
17 studies did not report offering referral services (at minimum, contact information for services that could be accessed, such as mental health hotlines) to participants, which is contrary to best ethical practice, particularly when measuring GBV (Ko-Ko-Zaw et al., 2011; Koçtürk et al., 2019; Mahmood et al., 2021; O. McCarthy et al., 2017). The lack of service referrals may have been a direct consequence of studies conducted during the COVID-19 pandemic when referral services were difficult to access or unavailable. Secondly, O. McCarthy et al. (2018) offered referral services only to female participants because collecting data on women’s contraceptive use in Tajikistan was deemed to be potentially sensitive, leading to compromised safety if women’s partners or parents were made aware. Support service information was not provided to male participants as the Tajik Family Planning Association advised it was not culturally realistic for males to feel unsafe after answering questions about contraception.
Studies that detailed the use of safety-related research practices were in the minority. However, when mentioned, practices were innovative. For example, Mahmood et al.’s (2021) GBV study mentioned that recruitment information was worded using general terms, in case abusive husbands saw the information or monitored internet access. Moreover, Abuhammah (2021a) did not force responses within their GBV survey in an effort to protect participants from negative feelings/distress. McCarthy’s (2017) SRH study incorporated instruction on how to delete or keep messages from the research team private. To monitor safety, the study also collected outcomes related to participants’ experiences of physical violence since enrolling in the study. Less than 1% of respondents had reported experiencing physical violence while being in the study (O. McCarthy et al., 2017, 2018). Other safety-related measures included asking participants if they shared a phone, reestablishing a different call back time, providing a nickname or alias, and choosing a time for participation that was convenient for them (Babalola et al., 2019; Mbulayi et al., 2021).
Seven studies explicitly mentioned participant literacy levels and only one study implemented measures to assist illiterate persons in participating (Mahmood et al., 2021). Researchers contacted illiterate or low literacy participants by phone and manually filled out the online questionnaire on their behalf. In other studies, inadequate literacy was noted as an exclusion criterion (Abuhammad, 2021a; Babalola et al., 2019; González-Hernández et al., 2020; O. McCarthy et al., 2017; Restar et al., 2020; Ybarra et al., 2020) and no reported measures were implemented to reduce participation barriers. While literacy levels may be less of a concern regarding telephonic data collection, illiteracy or low literacy may affect recruitment efforts that are contingent upon participants reading study information and opting in.
Lastly, considerations given to gender digital divide were not a predominant theme across the studies. Three studies (Mahmood et al., 2021; O. McCarthy et al., 2017; Ybarra et al., 2020) highlighted women’s more limited access to the internet and discussed strategies for mitigation. Women’s access to the internet also included affordability concerns for mobile data (Ybarra et al., 2020). To mitigate the gender digital divide, telephonic data collection was offered in lieu of the online survey (Mahmood et al., 2021), internet access was provided when unavailable (O. McCarthy et al., 2017), online surveys were compressed to reduce mobile data usage, and SMS surveys were used in lieu of online surveys (Ybarra et al., 2020). Lastly, Babalola et al. (2019) noted that husbands’ disapproval limited interactions with interventions intended to improve access to information about contraception.
Discussion
Critical Findings.
Implications for Practice, Policy, and Research.
The present review synthesized empirical, peer-reviewed research using RDC to measure any outcomes related to GBV and SRH in humanitarian and fragile settings for the purpose of better understanding and addressing methodological and ethical considerations of primary data collection concerning in a pandemic context. By improving methodological and ethical parameters of data collection related to GBV and SRH in these settings, we hope to ultimately expand the understanding of women and girls’ diverse experiences of violence and health. Overall, our synthesis indicates that conducting such research using RDC tools can maintain appropriate rigor and uphold ethical best practices under certain conditions. The rigorous and ethical application of RDC to collect sensitive information on SRH and GBV in humanitarian and fragile settings should be guided by best ethical and safety practices (Contreras-Urbina et al., 2019; Hossain & McAlpine, 2017; WHO, 2016) and may not always be appropriate in relation to the specified research goal.
Gap in the Evidence Base: Qualitative Research Using RDC Tools
Most of the studies included in this review utilized quantitative methods, indicating that the use of RDC tools to conduct qualitative research on GBV and SRH in fragile and humanitarian settings is even more limited in a pandemic context. The lack of qualitative research may pose epistemic challenges to collecting rich lived experiences from women and girls in humanitarian and fragile settings. Further, the ability to conduct realist evaluations of programs that incorporate qualitative data and triangulate methods may also be limited (Hossain & McAlpine, 2017).
Sampling and Response Rate Considerations
In humanitarian and fragile settings, it may not be financially or logistically viable to collect probability samples (Hossain & McAlpine, 2017), especially if the phenomenon under study is stigmatized. In such cases, researchers may use RDC tools (i.e., SMS, survey links) to facilitate peer-recruitment through respondent driven sampling. Moreover, while the use of online surveys disseminated through social media to collect prevalence data are popular, we advocate for a critical and thoughtful consideration of whether an appropriate sampling frame can be obtained. While prevalence data can help to prioritize needs, prevalence estimates are useful for planning if appropriate sampling strategies are used; otherwise, figures may over- or underestimate the true population prevalence parameter. Specifically, prevalence estimates using RDC tools may systematically exclude women and girls on the basis of inadequate internet access/affordability, low socioeconomic status, and suboptimal literacy. Accordingly, we reaffirm existing guidance for conducting research in humanitarian/fragile settings that prevalence data is not necessary for program development and can be considered a later-stage priority, particularly during periods of insecurity and infectious disease outbreaks (Hossain & McAlpine, 2017).
Regarding research specifically seeking to assess the feasibility or scalability of SRH or GBV services, promising ways to develop a sampling frame may include employing availability samples wherein all existing users of an application are invited to participate, using geo-located smart phone data, and screening potential respondents by demographic characteristics prior to accepting enrollment.
However, low response rates to survey questions disseminated among existing users of an SMS service or mobile application can introduce sampling bias if non-respondents are systematically different from respondents with respect to key variables. Thus, it is important to employ strategies that maximize participant retention such as disseminating survey questions immediately after users interact with the SMS service or mobile application, sending reminder messages using multiple modalities (SMS, instant message, email, phone call, etc.), and providing respondents with choice regarding amenable RDC tools.
Considerations for Measurement
Further, researchers must strive to incorporate validated and reliable SRH or GBV measures within RDC tools, taking into consideration that tools will need to be translated to the language of origin and back translated to ensure accuracy, pilot tested to assess psychometric properties, and adapted to the RDC tool selected. For instance, surveys delivered via SMS or instant messaging platforms could be simplified by using short versions of survey tools/scales/instruments. Further research is needed to develop and test publicly available short and reliable tools for measuring GBV and SRH that are intended for SMS or instant messaging platforms and appropriately normed among women and girls living in humanitarian settings. Such tools can be used in a pandemic context without sacrificing measurement quality. We also advocate for researchers to transparently report sample retention rates and make efforts to identify characteristics associated with non-response. This is of particular relevance among online surveys that are disseminated widely through social networking websites, thereby reaching persons who are eligible to participate but opt out. Given this review’s focus on primary data collection, all outcomes related to GBV and SRH were self-reported, and subject to social desirability bias, particularly if respondents were under perceived surveillance of family/friends.
Ethical Considerations
Regarding ethics, it is widely known that providing referral services is a well-established ethical best practice (World Health Organization, 2016). Researchers have a duty of care, particularly when conducting sensitive research in humanitarian and fragile settings wherein participants face unique vulnerabilities with respect to stigma, elevated stress, gender discrimination, and lack of legal protection from gender discrimination (Hossain & McAlpine, 2017). Thus, a prerequisite step to the use of RDC tools in humanitarian and fragile settings ought to be a grounded understanding of social socio-political norms pertaining to gender, SRH, and GBV in order to assess what referral services may be needed as well as acceptable and mapping out such services. Creating a community steering committee by using participatory approaches or collaborating and co-partnering with local Feminist organizations or advocacy groups can help provide valuable contextual insights into the social, cultural, safety and legal situation prior to and during data collection, particularly if the level of insecurity changes as the research is being implemented (Contreras-Urbina et al., 2019; Hossain & McAlpine, 2017; Lokot, 2019). Secondly, low literacy in combination with the gender digital divide may be of particular concern in fragile and humanitarian settings. Thus, researchers must incorporate strategies that mitigate barriers to participation among women/girls with low literacy and those affected by the gender digital divide, understanding that both may act concurrently to magnify barriers to participation in research employing RDC tools.
To advance research inclusion, it is imperative that participants be given different RDC tool options, depending on their literacy level and access to and familiarity with technology. Giving participants choice allows for greater freedom and opportunity to engage in research. However, the inclusion of women and girls in research by offering choice of RDC tool may come at a cost to measurement rigor; certain tools may introduce bias (i.e., social desirability bias may be greater threat in phone interviews compared to anonymous online surveys). Accordingly, researchers should collect data on what RDC tool was used and why in order to perform analytic control in regression analysis, sensitivity analysis, or hypothesize the direction of bias. Offering technological support is another strategy to mitigate participation barriers. Furthermore, of particular concern within humanitarian and fragile settings are inadequate internet infrastructure, affordability, and lack of compatible device ownership which can negatively affect respondent retention and participation; researchers should consider condensing internet-bases surveys for minimal data usage, using SMS for simple survey questions, and compensating participants for mobile data/internet charges.
The gender digital divide may manifest differently in terms of access, acceptability, digital literacy and internet access across humanitarian and fragile settings (Antonio & Tuffley, 2014). In humanitarian and fragile settings, normative barriers to women’s adoption of information and communication technologies include the perceived lack of relevance/need, negative perceptions of cost (particularly among male gatekeepers), fear of women and girls’ engagement in illicit relationships, infidelity, and exploitation, reputational damage, and beliefs that use of digital tools are a poor use of their time (Tyers et al., 2021; USAID, 2020). Thus, in collaboration with local groups, a context specific examination of the gender digital divide prior to data collection is prudent and appropriate measures must be incorporated to maximize women and girls’ participation in RDC.
The exclusion of participants who do not have adequate access to RDC tools can also introduce bias. For instance, research has shown that household ownership of computers, phones (fixed and mobile), and radio was associated with women rejecting patriarchal justifications of physical intimate partner violence (Cardoso & Sorenson, 2017). Computer ownership and increasing the number of owned information and communication technologies showed the strongest associations (Cardoso & Sorenson, 2017). Thus, by conditioning eligibility on technology access/ownership, studies investigating GBV may inadvertently be underestimating the true measures of association and prevalence.
Researchers may also consider incorporating specific safety features within RDC, such as asking participants if they have a safe phone number or email, establishing with participants a safe time to be contacted by research staff, introducing or describing the research purpose in broad terms (i.e., women’s health study), instructing participants to clear browser history or delete applications following study completion, and incorporating “quick escape” buttons (Seff et al., 2021).
Limitations
This systematic review has some limitations. First, due to the need for rapid evidence generation and synthesis, we could not accommodate two reviewers per title/abstract screened. Consequently, it is possible that the title/abstract screening yielded false negatives. We also did not consider non-English sources, which may have limited the sources considered for inclusion. Further, only considering peer-reviewed literature has the potential to omit relevant research conducted by monitoring and evaluation teams in humanitarian and fragile settings. Similarly, given the rapid expansion of remote data collection during the COVID-19 pandemic, it is possible that additional eligible studies were published between the database extractions and publication. However, we are confident that the range of included studies allowed for saturation of key methodological and ethical themes and best practice implications. Finally, we acknowledge that studies may have incorporated additional information related to our data extraction/charting, but simply did not report them.
Conclusion
This systematic review synthesized the existing evidence on the RDC of women and girls’ SRH and GBV within the context of humanitarian and fragile settings. Key methodological and ethical considerations were critically examined, including the consideration of an appropriate sampling frame when relying on online or mobile application samples, the introduction of bias when constraining eligibility on technology ownership/access, as well as how to minimize the gender digital divide and maximize women and girls’ safety during data collection. The COVID-19 pandemic and lack of vaccine equity in humanitarian and fragile settings pose considerable challenges to in-person data collection. This review is intended to aid SRH and GBV researchers in critically examining the appropriateness of using RDC in humanitarian and fragile setting.
Supplemental Material
Supplemental Material - Methodological and Ethical Implications of Using Remote Data Collection Tools to Measure Sexual and Reproductive Health and Gender-Based Violence Outcomes among Women and Girls in Humanitarian and Fragile Settings: A Mixed Methods Systematic Review of Peer-Reviewed Research
Supplemental Material for Methodological and Ethical Implications of Using Remote Data Collection Tools to Measure Sexual and Reproductive Health and Gender-Based Violence Outcomes among Women and Girls in Humanitarian and Fragile Settings: A Mixed Methods Systematic Review of Peer-Reviewed Research by Luissa Vahedi, Najat Qushua, Ilana Seff, Michelle Doering, Carrie Stoll, Susan A. Bartels, and Lindsay Stark in Trauma, Violence, & Abuse
Supplemental Material
Supplemental Material - Methodological and Ethical Implications of Using Remote Data Collection Tools to Measure Sexual and Reproductive Health and Gender-Based Violence Outcomes among Women and Girls in Humanitarian and Fragile Settings: A Mixed Methods Systematic Review of Peer-Reviewed Research
Supplemental Material for Methodological and Ethical Implications of Using Remote Data Collection Tools to Measure Sexual and Reproductive Health and Gender-Based Violence Outcomes among Women and Girls in Humanitarian and Fragile Settings: A Mixed Methods Systematic Review of Peer-Reviewed Research by Luissa Vahedi, Najat Qushua, Ilana Seff, Michelle Doering, Carrie Stoll, Susan A. Bartels, and Lindsay Stark in Trauma, Violence, & Abuse
Footnotes
Acknowledgements
The authors would like to acknowledge the supportive efforts of the project’s research assistants: Ms. Lily Coll, Ms. Anjali Gujral, Ms. Sonal Gupta, and Ms. Michelle Manivel. We recognize their supportive roles during screening and extraction.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the SSHRC Doctoral Fellowship (Social Sciences and Humanities Reserach Council of Canada).
Supplemental Material
Supplemental material for this article is available online.
Author Biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
