Objective 1. Understanding the impact of floods on healthcare utilisation, lives and livelihoods (March 2025–September 2025).

To understand the impacts of the floods on PLHIV, we will conduct semi-structured in-depth interviews (IDIs) and focus group discussions (FGDs) across the two study sites. Interview questions will cover topics around how PLHIV make sense of their experience of the April 2022 KZN floods – especially for the unexposed population, while the unexposed population (in Somkhele) will explore participants’ views about the 2022 KZN floods and other floods and EWEs experienced in their contexts. The interviews and FGDs will explore how PLHIV were impacted by the floods, including their health, healthcare access and utilisation and livelihood sustenance. We will also interview key informants from health and health-determining sectors, including the Department of Health, Department of Social Development, and Disaster Management Agency, to gain an understanding of actions or inactions towards mitigating the impact of the floods.

Objective 2. Measuring the economic losses and healthcare related costs (March 2025–December 2025).

We will follow the recommended methods to identify and evaluate the impacts of socioeconomic factors associated with floods and measure the economic impacts and losses, including healthcare utilisation and related costs, attributable to recent floods for a randomly selected sample of PLHIV. The study design, analytical methods and reporting of the findings will follow the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) Statement [43].

We will evaluate the economic burdens of floods in terms of direct and indirect costs (e.g., healthcare resources utilisation and work productivity loss). Individual-level data will be collected via structured questionnaires. A micro-costing approach (including a bottom-up construction of costs) will be employed to estimate costs associated with healthcare utilisation and other economic impacts of floods. The overall costs incurred due to floods will be estimated from a provider perspective (and a patient perspective, where data permits). We also collect socioeconomic and demographic information (e.g., age, gender, employment status, etc) of the study participants.

Objective 3. Co-development and co-prioritisation of policy actions (May 2025–May 2026).

In this objective, we will adopt a mixed-methods approach involving both quantitative and qualitative data collected through existing data sources, as well as engagements and workshops with key stakeholders involved in the care of PLHIV and those directly affected by the floods. This objective will be achieved through two main phases, as described below.

Phase 1: To identify early adaptive policy actions to minimise the health, economic and social impacts of floods on PLHIV

Phase 2: To create a priority list of early actions identified in Phase 1.

In phase 1, we will use the adapted Participatory Impact Pathways Analysis (PIPA) approach developed by the University of Sussex following the STEPS Centre [44,45]. PIPA provides a ‘road map’ that supports the co-development of a list of policy interventions by consulting and including multiple key stakeholders (e.g., patient representative, policymaker, health provider, etc). This approach is designed for complex contexts and provides a framework for ‘action research’ on processes of change. It has been successfully applied for co-production purposes in climate change research [45,46]. PIPA is a planning and, monitoring and evaluation tool designed to assist the people involved in a project, programme or organisation and make explicit their theories of change (i.e., how they see themselves achieving their goals and having an impact). Climate risk narratives are stories describing a subset of plausible – yet certainly not definitive – futures within the spread of climate projections intertwined with local context, impacts and vulnerabilities. Climate risk narratives enable the input of multi-model future climate scenarios into the project for exploring the health risks of future floods in PLHIV.

In phase 2, in line with the World Health Organization’s good practice for priority-setting [47–50], a multi-criteria decision analysis (MCDA)-based framework will be developed and implemented to rate, rank and prioritise the policy actions identified in phase 1 to support evidence-based climate change adaptation policymaking (particularly for flooding) involving key stakeholders and partners from all relevant sectors mentioned above via a workshop.

Objective 1. Understanding the impacts of recent floods.

Under objective 1, we will recruit a minimum of 30 participants (PLHIV) for the IDIs across the exposed and unexposed sites: about 15 participants per site to provide in-depth reflections on the floods and its impacts. We will also organise two FDGs – one in each site – consisting of about six to eight participants per group. We will aim to achieve age and gender balance in the selection of participants while taking into consideration the context-specific issues in each study site. Finally, we will recruit about ten key informants comprising policymakers from health, disaster management, social welfare and housing for the interview through the personal and professional networks of the research team. The actual number of IDIs and KIIs to be undertaken will depend on data saturation.

Objective 2. Measuring the economic losses and healthcare related costs.

The background loss from care after 12 months of antiretroviral treatment ranges from 20–30% [51]. Therefore, we will need to recruit a total of 424 PLHIV (212 unexposed and 212 exposed individuals) at a matched exposed-unexposed ratio of 1:1 (one unexposed for one exposed) to reach an 80% power at 0.05 significance level to detect a 90% increase in odds of loss from care during the first 6 months following the floods. Although we do not aim to estimate loss from care (i.e., loss from follow-up) in this study, it underpins our study hypothesis while not the primary reason for this study. The AHRI HDSS databases described in the participant recruitment section will serve as the sampling frame for randomly selecting the 212 unexposed individuals in Hlabisa, of whom 70% will be female, which is representative of the proportion of females amongst all PLHIV who engage with the surveillance programme. A similar approach will be taken to recruit 212 exposed PLHIV in Ndwedwe by working closely with AHRI CED, community caregivers and other community-based organisations, as mentioned above.

Objective 3. Co-development and co-prioritisation of policy actions.

In this objective, we will engage a minimum of at least 15–20 stakeholders, including policymakers in health and health-determining sectors (water, housing, energy), civil society and advocacy groups, healthcare providers, PLHIV, and health system researchers who will be invited through the personal and professional networks of the research team in the workshops in both phases of the objective.

Objective 1. Understanding the impacts of floods.

IDIs and KIIs are expected to last between 45 minutes and 1 hour. The FGDs will also last between one to two hours. The duration of these interviews often depends on the disposition of participants and their willingness to actively engage (or not). It thus requires the interviewers’ ability to communicate and engage effectively with the participants. Consequently, data collection under this objective will be undertaken by experienced social science researchers.

All interviews will be audio-recorded, transcribed verbatim and translated to the closest English meaning by experienced social science researchers. Where a participant refuses to be audio-recorded for the interviews, the interviewer(s) will take adequate notes – mental and physical – to help with the development of a comprehensive summary narrative that would suffice in lieu of the interview transcript [52]. The note will endeavour to comprehensively capture what was said by the participant(s). These transcripts will be quality controlled by trained social science study coordinators to ensure their quality for data analysis. The data collection for this objective commenced on the 4^th of April 2025 in Hlabisa (the unexposed study area). All recruited participants provided written informed consent, which was documented by the local research team.

Objective 2. Measuring the economic losses and healthcare related costs.

The primary data for this objective will be collected after the preliminary analysis of qualitative data collected for objective 1, which will feed back into reviewing and finalising the questionnaires for objective 2. Study questionnaires will then be pilot-tested with a small portion of the study sample to identify potential issues and assess the rigour of questionnaires. The findings of the pilot study will be analysed to inform, test and revise our econometric model. The questionnaire for this objective has been initially developed by the research team (which includes a clinician, climate scientist, social scientist, health economist and policy analyst) and was then pre-tested by interviewing six researchers from a variety of backgrounds (e.g., ethics, anthropology, global health sciences, nursing, economics), as well as two operational managers from Somkhele and Ndwedwe community health clinics and two research assistants experienced in working with PLHIV in rural South Africa. This allowed to check for clarity, understandability and relevance of the questions and revise the questionnaire. The pre-tested questionnaire is expected to take 60–90 minutes to complete.

Trained supervised research assistants will facilitate and make appointments with participants to complete questionnaires. Face-to-face interviews will be carried out one by one with a research assistant in one of AHRI Somkhele office’s rooms (for the unexposed group) and the Ndwedwe Community Health Centre or a participant-researcher mutually agreed location (for the exposed group) who will use a tablet to enter data into the Research Electronic Data Capture (REDCap), a secure web-based platform widely used for data collection and management for this objective. The coordinating team at AHRI will export data from REDCap for data curation and cleansing.

Objective 3. Co-development and co-prioritisation of policy actions.

Phase 1 involves using the core elements of PIPA that include the problem tree analysis; visioning exercises; current and future network maps, indicating interactions and the linkages required to achieve project aims; and the ‘outcomes logic model’ [53,54]. Different groups of key stakeholders will be involved, and the PIPA approach will be employed through a workshop in one of AHRI offices [44,54] for their framings of how they see PLHIV and their healthcare being affected by floods, now and into the future. This follows the recognition that who you are shapes how you frame or understand a system or a problem. The information from this phase will be used to identify adaptation early action interventions. For instance, PLHIV, a local health clinic worker, a hospital doctor, a climate scientist, a national health minister, and a multinational pharmaceutical firm might all frame the climate issues affecting PLHIV in different ways. Those various framings will lead to different narratives being told about the same system and different choices being made. Too often the narratives of powerful actors and institutions become the motorways channelling policy, governance and interventions, overrunning the valuable pathways responding to poorer people’s own goals, knowledge and values.

The pathways approach examines multiple pathways and backed by a variety of practical methods that opens up space for more plural and dynamic healthcare resilience [44,45]. It aims to open up the political process of building pathways that are currently hidden, obscured or oppressed. Specific practical methods used in this project will include system mapping of climate-health links that are relevant to PLHIV, PIPA and climate risk narratives [55]. Systems mapping of climate-health links helps make explicit and visualise the causal relationships and pathways from floods to poor livelihood and healthcare outcomes [56].

Phase 2 of this objective also involves two other workshops to be held in the AHRI offices, engaging the same groups of key stakeholders mentioned above in phase 1. In workshop 1, an overarching set of prioritisation criteria will be specified through a brainstorming and think-aloud session of stakeholders, including PLHIV and public advocates. Workshop 2 involves the priority-setting exercise, where actions identified in phase 1 will be co-rated, co-ranked and co-prioritised based on criteria finalised in workshop 1. The priority list of actions will be simplified into three tiers of priority: i.e., ‘high urgency’, ‘medium urgency’ and ‘low urgency’ for ease of communication (e.g., with researchers and policymakers) [57].

Objective 1. Understanding the impact of recent floods.

We will analyse the data from the IDIs, KIIs and FGDs that have been transcribed, fully translated and quality-controlled using the thematic analysis approach [58,59]. Field notes and or interview summaries will also serve as data sources with the translated transcripts to give a full picture of the data from the purview of the interviewer(s)/researchers. The data will be managed through the NVivo data analysis software, and the coding process will be a mix of inductive and deductive thematic coding. We will however also carry out an initial open coding of individual manuscripts to ascertain general codes emerging within the data. This will then be followed by organising them into themes and subthemes based on some pre-identified categories linked to the social, economic, health and mental/psychosocial impacts of floods, including adaptation and mitigation strategies. Field notes and thematic summaries will be developed and prepared on the key topics as they emerge from the interviews. Preliminary findings will be drawn from summaries to aid in the preparation of briefing notes to disseminate these findings among our stakeholders.

Objective 2. Measuring the economic losses and healthcare related costs.

Following the recommended methods, we will construct a multi-variate econometric model to identify, understand and measure the impact and extent of socioeconomic impacts of recent floods across PLHIV, adjusting for baseline covariates such as age, gender, etc. We will also conduct a cost analysis to measure the economic burdens of floods in terms of direct and indirect costs in South African Rand (SAR), local currency. The SAR will be then converted to US dollar (USD) using the purchasing power parity (PPP) conversion factor recommended by the World Bank [43,60]. The economic impacts of the floods, in terms of the loss of work productivity, will be estimated using the human capital approach to reflect the value of lost production due to absenteeism and presentism associated with recent floods for PLHIV [61,62]. Both descriptive and inferential statistical analysis will be used for comparison purposes, as well as sub-group analysis across exposed and unexposed groups. Following recommended methods, multiple imputations will be used to deal with missing data [63–65]. We will calculate confidence intervals and undertake simulation, bootstrapping and sensitivity analysis to map and assess the uncertainty in our parameter estimates and the robustness of our findings. We will also undertake an equity impact analysis through subgroup analyses of the economic impacts of floods linked with the socioeconomic and demographic characteristics (e.g., age, gender, wealth quintile, etc) of the study sample across exposed and unexposed groups. Data analysis will be undertaken in STATA 16.

Objective 3. Co-development and co-prioritisation of policy actions.

After the Phase 1 workshop, the facilitators will use the workshop outputs to synthesise the outputs qualitatively and quantitatively, prepare narratives including the underlying logics and assumptions and construct the outcomes logic model and the network maps. The workshop outputs will be visualised using MS Office tools or one of the social analysis software available for free [43–45].

Phase 2 consists of developing a MCDA-based framework in MS Excel for prioritising policy actions and interventions that will involve the following three key components: (i) identifying policy actions (output of phase 1) to be prioritised; (ii) specifying prioritisation criteria (i.e., the characteristics or attributes of actions); and (iii) measuring weights for the criteria, representing their relative importance to stakeholders (e.g., patient and public advocates, researchers, policymakers, etc) [49,50]. These components will be applied to co-rate each policy action according to its ‘performance’ on each criterion. Finally, the ratings will be aggregated using a weighted sum model – to co-produce a total score (typically in the range of 0–100%) for each action [49,50,57,66], by which the actions are ranked (and prioritised). A sensitivity analysis will be performed to check the robustness of the results.