Healthcare system resilience in Bangladesh and Haiti in times of global changes (climate-related events, migration and Covid-19): an interdisciplinary mixed method research protocol

According to the Global Climate Risk Index in date, Bangladesh was ranked 7th and Haiti 3rd on the Long-Term Climate Risk Index [20]. Both countries are prone to disasters due to their geographical location and environmental characteristics. They are subject to hydro-meteorological disasters; especially cyclones, heavy rainfalls and flooding. Besides, Haiti is facing geophysical hazards. Vulnerability to climate change is also related to the combined consequences of other structural and institutional factors: deep social inequality, high poverty rate, poor public health infrastructure and accessibility, land use and access, high population density (Bangladesh), rapid urbanisation, civil unrest and violence (Haiti) [21‐23]. The context in Haiti is that of severe socioeconomic deterioration, denunciation of corruption, long lasting political crisis and recent constitutional crisis. This has resulted in lockdowns, massive social unrest and demonstrations. Since 2018, Haiti has been experiencing a significant increase in human rights violations and abuses, crime and massacres, kidnappings, restrictions on freedom of expression and freedom of the press, continuing impunity and aggravated food insecurity [24, 25]. This serious socio-political instability has led to the closure of some care facilities [26]. Mobility, whether climate-induced or not [27], could put an additional strain on healthcare systems. Because population mobility and migration have multiple causes and have a complex relationship with climate change, scientists are cautious about linking mobility and migration to climate change [28]. However, some climate-related events have been shown to directly or indirectly influence mobility and migration decisions [29]. This nuance being stated, mobility and migration affect the healthcare system at two levels: at the organisational level, for example with a structure, services and workforce that are not adapted to the patients and at the population level with an increased vulnerability of the population. Both health system performances are poor: Bangladesh is ranked 88th and Haiti 148th out of 191 countries in 2000 by the WHO. When considering the UHC effective coverage index in 2019,Bangladesh is in the 6–7 decile and Haiti is in the lowest decile [30].

Exploratory key informant interviews and reviewed secondary data resources were conducted to select two sites in each country, based on climate-induced vulnerabilities, mobility patterns and availability of health system secondary data for the past decade when possible. Sites boundaries are the ones defined by local health system units. We selected for their local context two rural coastal Haitian sites (Anse d’Hainault and Môle St Nicolas) and one rural (Tala) and one urban (Duaripara slum) Bangladeshi sites (Fig. 2, Table 1).

Regarding migration and mobility in Bangladesh in 2020, 88,000 people living in displacement (IDPs) were registered due to disasters and 427,000 IDPs due to conflict and violence, which, for most, are protracted cases from the partitioning and independence of the country, were recorded [31]. As of 31 December 2019, 4,086,000 new displacements due to disasters were recorded, many in the form of life-saving pre-emptive evacuations, and 520 new displacements due to Intercommunal violence against Ahmadi Muslims in Rangpur division and political violence following elections [31].

Tala upazila (or Tala subdistrict) is located in the south-western coastal region of Bangladesh, in the Satkhira district of Khulna division, which is one of the most disaster-prone areas of the country, especially for flooding and waterlogging [32, 33]. Tala has at least 30,000 inhabitants, 15 to 20 community clinics and five sub-districts hospitals. The primary source of income is agriculture. Men, pushed out by lands waterlogging, migrate out for labour, while women, children and elderly, remain behind with poor access to sanitation and drinking water. Pregnant women often suffer from reproductive health issues because they cannot access healthcare [34].

Duaripara slum is located in ward 6 of Dhaka North City Corporation. Most households are extremely poor according to BRAC Urban Development Program (UDP). Its population increased recently when residents from the nearby Bhola slum moved here after they were evicted. Most of the Bhola slum residents migrated from Bhola Island after losing their land due to riverbank erosion or property loss and damage due to cyclones [35]. Healthcare is mostly limited to over-the-counter treatment at pharmacies; residents go to government and private hospitals, which are far, only during serious illnesses. 24.7% of the residents work in a garments factory followed by 15.6% of residents who work as day labourers and other significant occupations are rickshaw-pulling and domestic work [36]. All houses have electricity, although non officially, and piped water is shared by around ten households. The slum settlement area is built on wetlands. Flooding, standing water, fires, and storms, especially during the monsoon season, are frequent in the area.

In Haiti, many people displaced by the 2010 earthquake and hurricanes such as Matthew in 2016, are still living in displacement [37]. As of 31 December 2019, the total number of people living in displacement (IDPs) includes 51,000 IDPs due to disasters and 2100 IDPs due to conflict or violence [37]. For 2019, 1200 new displacements due to disasters and 2100 due to conflict and violence were recorded in Haiti [37].

Anse d’Hainault arrondissement is located in the Grand’Anse Department which has been affected by Hurricane Matthew in 2016. It has an estimated population of 98,522 inhabitants (IHSI estimates for 2015), with an extreme poverty rate of around 35% and counts 11 health facilities: six dispensaries, two health centres without beds, one with beds and two community hospitals. Coastal plains are subject to marine submersion as well as floods. Essential services like access to water or electricity are deficient. In rural areas, as in Mole St Nicolas, agriculture is the primary source of income.

Môle Saint-Nicolas arrondissement has an estimated population of 245,590 inhabitants (IHSI estimates for 2015) and is considered one of Haiti’s poorest departments, with an extreme poverty rate above 40%. It counts 36 health facilities: 30 dispensaries, two health centres without beds, three with beds and one community hospital.

We use a mixed-methods sequential exploratory research design [38] with a four-phase organisation (Fig. 3: framework design).

The first exploratory (contextual) phase consists of a contextual analysis to describe, political, social, economic, and health factors and their relationships, both at the national and local levels, through 1) a compilation of local monographs describing the four sites’ socio-political organisation, the history of the communities and their livelihoods, local healthcare system and demographic trends, 2) public health and public health reforms and migration policies at the national level, and any other experiences that have an impact on the local health system and population, through interviews, document analysis, and public service records and archives. In addition, a quantitative analysis of secondary data from routinely collected data (DHMIS) and SARA surveys, combined with the development of indicators, will be conducted to better understand the impacts of risks, events and climate change over the past decade on healthcare service providers (HSPs) at the four sites. These analyses are intended to allow for longitudinal and transversal analyses.

For the second phase, we use a qualitative methodology with a mixed inductive-deductive approach. This phase aims to map out the variety of disruptions and disturbances faced by HSPs and the populations and identify the diverse coping strategies implemented to understand the factors and stages in the process of resilience in the four sites. Mobility and migration will be part of the discussion on coping strategies of the population. In-depth and semi-directed interviews with national (n = 2) and regional/local (n = 1 + 2) health decisions makers and with the heads of the 10 selected healthcare service providers within each site, and focus group discussions with inhabitants of the community (n = 3), community-based groups (n = 3) and healthcare users (n = 3) will be organised within each site too. Those numbers will be adapted following a purposeful sampling strategy, based on criteria (age, sex, socio-economic status, being a patient, place of living) and according to the results of pilots and each previous phase [39]. Finally, a mental health research arm, whose specific protocol will be described in another publication at a later stage, will be carried out to further investigate issues pertaining to mental health and trauma in the face of extreme environmental events. For this embedded study, qualitative data will be collected to complement our quantitative data, at a different timing than phase 2.

During the third phase, all HSP health staff and a sample of household members from the communities around the 10 selected HSPs will be interviewed to complete questionnaires (n = 500 households per site- see below for sampling strategy and eligibility criteria). This quantitative phase aims to quantify the (perceived) impacts of identified disruptive events, changes and risks on the healthcare providers, healthcare access and population health status, based on the results from phase 2. Specifically, the household questionnaire includes different modules in two sections, a household and a respondent section. The respondent section includes socio-demographic information, health status (acute and chronic diseases), reproductive health, maternal health and family planning, mental health and trauma, empowerment, history of health and history of mobility, social capital, livelihood security and community perception, and a module on healthcare access. The household section includes household socio-demographic and economic information. The HSP questionnaire includes questions on five supply-side determinants on healthcare access [15] in connection with the different SARA surveys available in both countries. It will describe the situation and assess, if possible, the impacts of disruptive events on local healthcare service providers. Interview guides (phase 2) and questionnaires (phase 3) are presented in the supplementary materials.

The fourth phase aims to document further and better understand the typologies and strategies that will have emerged from the two previous phases. During this qualitative phase, a nested sample of health professionals (n = 20) and household members (n = 30 households) will be interviewed.

We will select ten HSPs per site (20 per country) to represent the diversity of levels (first and second), geographical situations (urban and rural) and types (private, for-profit, private non-profit, public) of each site. We will consider only HSPs with a minimum attendance level: five consultations per day in Haiti and 30 in Bangladesh.

For the phase 2, we will identify key informants who will be able to provide specific information on our study topic [39] and will be selected to have a wide range of understanding of the phenomenon. Regarding the decision-makers (n = 5), they will be interviewed at the local (n = 2), regional (n = 1) and also national levels (n = 2).

Community inhabitants (mostly non-users of healthcare), community groups (women’s groups, trade unions, community health workers, etc.) and users of healthcare services (15 patients will be randomly selected from three of the 10 healthcare service providers selected above, while taking account of gender and residence distance from HSP), will be enrolled for focus groups participation according to their knowledge and experience. Community leaders, local NGOs and healthcare facility managers will assist us in selecting these individuals.

To implement phase 3, 500 households per site (1000/country), spread over 10 clusters corresponding to the 10 HSPS will be selected in each site. Clusters will coincide with circles with a kilometre radius in Bangladesh and a 1-to-3 km radius in Haiti around the previously selected HSPs. In each circle, we will identify households through photo interpretation. Then, we will select a spatialized random sample of 50 households.

The household survey will be conducted to select households that are

Whenever possible, we will interview all adults who slept in the home the previous night. Questions about children and under 18 who also slept in the home the previous night will be asked of the child’s primary caregivers. A minimum of 100 adults per cluster will be interviewed. The number of households visited will be adjusted accordingly but should be around 50. Finally, during Phase 3, all health staff of the 10 local HSPs will be administered the HSP questionnaire.

For the qualitative phase 4, a nested sub-sample of households and HSP will be selected from among those who responded to the Healthcare Service Provider and Household Questionnaires in Phase 3, for in-depth qualitative interviews with HSPs (n = 16) and users and non-users (n = 40). These numbers are approximate as they will be determined based on the preliminary results of the quantitative analyses.

We will recruit a sample of 500 households per site (N = 4 × 500). Sample size calculations are based on the main outcome variable, scale for self-rated health [40]. Our targeted sample of 2000 households would yield a minimal detectable size of 12% at 80% power for a type I error (α) of 5% [41] using a linear regression. Such effect size is comparable or below to estimates reported in the literature [42, 43].

A research data management plan will be created to help researchers manage the data as part of their research activity or project as well as to facilitate the sharing of data at the end of the project at https://​dataverse.​ird.​fr.

Ethics approvals have been granted from the Institutional Review Board (IRB) of the BRAC James P Grant School of Public Health, BRAC University (ref: IRB-19 November’20–050) in Bangladesh and from the Comité National de Bioéthique in Haiti (ref: 2021–10). We will not collect biological data. We will ensure usual considerations concerning fieldwork ethics and research involving humans. However, in Haiti, given the complex socio-political context, an advisory committee including national researchers, civil society representatives and health system focal point from both selected sites, will be set up to discuss issues of health ethics principles, climate change [48] and violence, in particular. Besides, this committee will provide advice and guidance in data collection in difficult areas. If conditions for quantitative data collection become too challenging, a new protocol, involving only qualitative research and a maximum of online/telephonic qualitative interviews, will be proposed. It will include the qualitative phase described here (phase 2), followed by another qualitative phase (new phase 3). This updated phase 3, the design of which will depend on the results of phase 2, will establish a typology of events and individual, household and community strategies when facing the events, since the phase 2. If necessary and possible, a third qualitative phase following similar patterns will follow.

Participation to the study will be voluntary. We will inform participants about the research results’ intended publications, information on data use and sharing. We will also inform the participants about their right to withdraw at any time without consequences and about their right to refuse to answer. For the quantitative data collection, the consent form will be completed for all participants or those legally allowed to consent on their behalf. Because being underaged, the consent forms will be asked to the adolescents and their caregivers. Questions about children will be asked to their caregivers. We will interview participants in an adequate setting allowing confidentiality. We will replace the respondents’ names by ID codes that will not appear in the public databases. Once all data are imputed and ready to use, the key, associating names and ID codes, will be destroyed. For the other data collection, depending on the people interviewed and the local issues at stake, written or verbal consent will be systematically requested from participants. The usual ethical criteria for qualitative health research will be respected. Methodology concerning consents was approved by both ethics’ committees in Bangladesh and Haiti. A pilot study will assess the cultural sensitivity of some questions and adjustments will be decided with the local teams. Most of the sensitive questions have already been tested for the SARA/DHS surveys. Enumerators are trained to appropriate policy and practice concerning confidentiality, anonymity or acknowledgement of research participants. Feedback to participants on the research results as appropriate will be planned.

Knowledge transfer will take place throughout the project, in close collaboration with stakeholders. Support will be offered in each of the two countries to develop and implement a knowledge transfer plan. These plans will be adjusted throughout the project according to its evolution. The plans will (at least) include the presentation of results through research snapshots and policy briefs [49] in English and national languages. We will also disseminate our finding at the national, regional and local Health Authorities through deliberative workshops [50] informal meetings, presentations, and policy briefs, which will have been reviewed by stakeholders for content and presentations [49] and also to other countries struggling with climate change, through policy briefs as well as through international and regional networks. Scientific communication is planned through publications and presentations at national and international conferences. Part of the material, such as research snapshots, policy briefs, evaluation protocols, videos and evaluation findings will be publicly available (after publication for any results dissemination) through the project website.

Participatory visual anthropology will also be part of our knowledge transfer strategy. According to Baumann’s methodology [51], we will facilitate the creation of participatory videos in different communities. These videos will visually represent the complex issues of migrant access to healthcare, advocate for an embodied and bottom-up approach to healthcare issues and help communities to dialogue with public decision-makers. We will develop a Memorandum of Understanding with project participants to include project objectives based on stakeholders and consent to film and/or be filmed. In addition, a workshop will be organised between the viewing of the edited film and the broadcast to obtain consent. Consent will also be the subject of in-depth discussions on image rights, broadcasting rights and signing of a form by participants and researchers. A specific research protocol for these videos will be published.