Enhancing epidemic forecast usability for policymakers: A global mixed-methods study
Source: PubMed Central Open Access, NCBI / U.S. National Library of Medicine
The COVID-19 pandemic exposed critical gaps in the generation, interpretation, and use of epidemic forecasts for public health decision-making. We conducted a global mixed-methods study combining an online survey (n = 143, from 46 countries across all World Bank income groups) with 13 semi-structured interviews to examine how epidemic forecasts were perceived, used, and communicated by stakeholders involved in COVID-19 policy dialogues. Survey responses were analysed descriptively, stratified by country income group, while interview transcripts were analysed thematically using the Framework Method. Forecasts informed policy questions ranging from epidemic size estimation to intervention planning, with the projected impact of interventions (65%), epidemic peak (64%), and prevalence (62%) being the most frequently communicated metrics. Preferred formats varied by setting: 72% of high-income country (HIC) respondents valued explicit uncertainty presentation, compared with 34% in lower-middle-income countries (LMICs) and 23% in low-income countries (LICs). Barriers to forecast use were most pronounced in lower-income settings, where 47% of LIC and LMIC respondents reported that colleagues did not understand the modelling methodology, compared with 3% in HICs. Qualitative data highlighted that forecast credibility depended on interpersonal trust, institutional relationships, and contextual relevance rather than statistical sophistication alone. Findings should be interpreted in li
Abstract
The COVID-19 pandemic exposed critical gaps in the generation, interpretation, and use of epidemic forecasts for public health decision-making. We conducted a global mixed-methods study combining an online survey (n = 143, from 46 countries across all World Bank income groups) with 13 semi-structured interviews to examine how epidemic forecasts were perceived, used, and communicated by stakeholders involved in COVID-19 policy dialogues. Survey responses were analysed descriptively, stratified by country income group, while interview transcripts were analysed thematically using the Framework Method. Forecasts informed policy questions ranging from epidemic size estimation to intervention planning, with the projected impact of interventions (65%), epidemic peak (64%), and prevalence (62%) being the most frequently communicated metrics. Preferred formats varied by setting: 72% of high-income country (HIC) respondents valued explicit uncertainty presentation, compared with 34% in lower-middle-income countries (LMICs) and 23% in low-income countries (LICs). Barriers to forecast use were most pronounced in lower-income settings, where 47% of LIC and LMIC respondents reported that colleagues did not understand the modelling methodology, compared with 3% in HICs. Qualitative data highlighted that forecast credibility depended on interpersonal trust, institutional relationships, and contextual relevance rather than statistical sophistication alone. Findings should be interpreted in light of potential recall and selection biases inherent in retrospective, convenience-sampled designs. Strengthening forecast impact will require modular, user-oriented tools, embedding modellers within response teams, co-developing decision-relevant metrics, and sustained investment in foundational health information systems, particularly in resource-constrained settings.