Implementing a context-driven awareness programme addressing household air pollution and tobacco: a FRESH AIR study

Chronic respiratory diseases impose a major global health burden, disproportionately affecting low- and middle-income countries where risk factors such as tobacco use and household air pollution are prevalent and resources to address them are limited. Previous interventions often fail due to misalignment with local knowledge, beliefs, and context, resulting in low motivation for behavior change. An awareness and empowerment programme developed in Uganda, grounded in the COM-B model and using a cascading train-the-trainer design (HCWs to CHWs to communities), showed feasibility, acceptability, and effectiveness. The research question was whether this programme could be feasibly, acceptably, and effectively translated and implemented in two very different low-resource contexts—Kyrgyzstan and Vietnam—by integrating it into existing health infrastructures and co-developing with local stakeholders.

The paper frames the problem with evidence that CRD burden and associated risk factors are high in LMICs, and that implementation failures often stem from poor contextual fit. The COM-B model underpins behavior change efforts by targeting capability, opportunity, and motivation. Prior programmes in LMICs often focused on tobacco alone, with growing recognition of the need to address household air pollution. Studies highlight that embedding interventions into existing health systems, engaging CHWs, stakeholder involvement, and compatibility with local context are crucial to success. Clean cooking interventions frequently report low adoption unless contextual barriers are addressed. The Ugandan precursor programme demonstrated promise, motivating exploration of transferability to other settings.

Study design: Prospective implementation study (2016–2018) within the FRESH AIR project, reported per StaRI standards. The programme and implementation strategy are detailed in Box 1 and Fig. 1. Setting: Kyrgyzstan (high prevalence of solid fuel use, tobacco consumption 26%) and Long An province, Vietnam (75% solid fuel use; tobacco 23%). Pre-FRESH AIR work indicated poor CRD awareness. Sites were chosen based on opportunity and existing relationships. Population: Eligible participants included HCWs, CHWs, and community members without additional inclusion/exclusion criteria. Initial HCWs were selected with local stakeholders (e.g., district health officers). These HCWs then conveniently selected further HCWs/CHWs, who then trained nearly all community members in their villages. Intervention: A cascading train-the-trainer awareness programme to increase knowledge on CRDs and risks from tobacco and biomass smoke, and to empower behavior change. Content covered smoking cessation, second-hand smoke reduction, and HAP reduction via source (clean fuels, improved stoves), environment (ventilation, kitchen design), and user practices (drying fuel, pot lids, stove maintenance, keeping children/pregnant women away, outdoor cooking). Initial 3-day HCW training included content co-creation, training skills, and community mobilization techniques; then HCWs trained other HCWs (1 day), who trained CHWs (half day), who trained communities. Materials (posters, flip charts, brochures) were translated/adapted to local context with stakeholder co-creation and approved by national authorities. The implementation strategy was co-developed with local influential stakeholders to ensure embedding in existing infrastructures. Outcomes: Feasibility (implementation within reasonable effort, budget, time), acceptability (positive responses by deliverers/recipients), fidelity (adherence to planned steps), and effectiveness at multiple levels: short-term knowledge (pre/post questionnaires for HCWs, CHWs, community), and longer-term behavioral indicators via acceptability and adequate use (stove stacking) of improved cookstoves distributed in a subsequent FRESH AIR programme. Data collection: Ongoing discussions and consensus among teams captured feasibility, acceptability, and lessons learned. Knowledge questionnaires contained true/false/I-don’t-know items, translated/back-translated to Russian/Vietnamese. Improved stove acceptability assessed via questionnaires; adequate use inferred from observed stove stacking. Analysis: Qualitative analysis of feasibility/acceptability guided by the modified Conceptual Framework for Implementation Fidelity. Knowledge changes analyzed with Wilcoxon signed-rank tests (p<0.05). Behavioral indicators summarized descriptively. Sample size: Targeted ~400 community pre/post questionnaires (budget-limited). For stove outcomes, 20 households in Kyrgyzstan and 76 in Vietnam were randomly invited. Ethics: Approved by ethics boards in Vietnam (188/DHYD-HD; 06/27/2016) and Kyrgyzstan (5; 03/03/2016). Written informed consent obtained for stove participants; other activities fell within routine roles or voluntary attendance of education.

• Feasibility and acceptability: High in both settings; implementation completed within the 3-year project timeline and within intervention budgets of approximately €11,000 per country (Kyrgyzstan: €10,900; Vietnam: €10,050). Including study activities, totals were €15,400 (Kyrgyzstan) and €15,000 (Vietnam).
• Reach (coverage): Kyrgyzstan—initial 10 HCWs trained; ultimately 90 health workers trained (vs. planned 50), each reporting contact with 80–160 community members monthly; ~15,000 community members reached within 6 months. Vietnam—17 HCWs initially trained (one per health center), leading to 77 trained CHWs; each CHW contacted 100–150 community members; ~10,000 individuals reached within 6 months.
• Fidelity: Most steps adhered to as intended, including co-development with stakeholders, co-creation of materials, and cascade training. Minor adaptation in Kyrgyzstan due to long travel times leading to adjustments in cascade structure.
• Knowledge gains: Significant pre- to post-training increases across all groups in both countries (Wilcoxon p<0.05). Baseline knowledge lower in Kyrgyzstan with larger improvements.
• Behavioral indicators (improved cookstoves): Acceptability high—100% of stove users in Kyrgyzstan and 89.8% in Vietnam would recommend the new stove. Stove stacking: 15% in Kyrgyzstan versus 85.5% in Vietnam. In Vietnam, many considered the improved cookstove too small; 44% continued daily use of traditional stoves and 36% used them several times a week.
• Determinants of success: Essential components included deep contextual knowledge and embedding in existing health infrastructures (notably leveraging CHWs), early and ongoing engagement of influential local stakeholders to build ownership, and flexibility to adapt to emergent contextual factors.
• Media and training quality: Kyrgyzstan used national TV, radio, and newspapers for awareness; both sites emphasized training-of-trainers, used PowerPoint/flip-overs, and provided refresher courses where budget allowed.
• Costs: Detailed intervention and study costs reported; local variations included higher travel costs in mountainous Kyrgyzstan and mandatory compensation for training time in Vietnam.

The study demonstrates that a contextually adapted, cascading train-the-trainer awareness programme on CRD risks can be feasibly and acceptably implemented in disparate low-resource settings when embedded in local health systems and co-developed with stakeholders. Significant knowledge gains indicate effective enhancement of psychological capability (COM-B), which is a prerequisite for behavior change. Differences between settings—greater knowledge gains and higher stove adoption in Kyrgyzstan versus higher stove stacking in Vietnam—highlight the influence of baseline awareness, prior traditions of patient education, and technology fit (e.g., stove size adequacy) on behavioral outcomes. The findings reinforce the centrality of CHWs in delivering preventive interventions, the importance of contextual compatibility and stakeholder ownership, and the need for flexible adaptation during implementation. While clean cooking behavior is multifactorial and requires supportive markets and policies, the programme provides a scalable approach that can align multiple actors (policy makers, health workers, communities) to address determinants simultaneously. Reporting on implementation fidelity and costs strengthens interpretability and relevance for scale-up across LMICs.

Contextual translation of an evidence-based, train-the-trainer awareness programme from Uganda to Kyrgyzstan and Vietnam was feasible, acceptable, and effective in increasing lung health awareness and empowering communities to reduce exposure to biomass and tobacco smoke. Embedding within existing health infrastructures, engaging influential stakeholders to build ownership, and maintaining flexibility were critical to success. The approach offers a practical guide for adapting and implementing lung health interventions in diverse low-resource settings and can serve as a foundation for complementary programmes (e.g., smoking cessation, clean cooking). Future work should include controlled designs to assess causal impacts on long-term behaviors and health outcomes, validate knowledge instruments, optimize technology fit (e.g., stove characteristics), and integrate supportive policies and market mechanisms to sustain behavior change.

• Reliance on self-reported implementation integrity (numbers of sessions/participants), as budget did not allow comprehensive in vivo observation; potential for social desirability bias.
• Knowledge questionnaires were not previously validated; potential selection bias in respondents.
• No control group; causal attribution of longer-term behavioral outcomes (stove acceptance/use) to the awareness programme is uncertain.
• Financial incentives/compensation (e.g., stove provision or cost coverage) may have reduced typical financial barriers, limiting generalizability to routine settings.
• Miscommunication led to a costly pilot study in Vietnam not initially planned; also, bureaucratic approvals caused delays though necessary for sustainability.
• Some deviations in cascade structure (e.g., in Kyrgyzstan due to travel constraints) and incomplete tracking of dropouts in Vietnam.