Bridging the Gap: Enhancing the Evaluation & Interpretation of Epidemic Forecasts for Researchers & Policymakers in Resource-Constrained Settings
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
The COVID-19 pandemic highlighted the complexities of crisis decision-making, where evidence generation, interpretation, and action occurred simultaneously. These processes unfolded under compressed timeframes, uncertainty, and politically charged environments. Epidemic forecasts became key tools for navigating uncertainty, yet their use varied widely. This study aimed to explore how forecasts were used, interpreted, and communicated in pandemic decision-making, and how knowledge translation could be improved.
Methods
We conducted a global mixed-methods study, combining an online survey and semi-structured interviews with individuals involved in COVID-19 policy dialogues across 46 countries. Eligible participants had directly or indirectly contributed to pandemic policy decisions. Of 143 survey respondents, 105 (73·4%) reported involvement at national, subnational, or field levels. We conducted 13 interviews with individuals engaged in COVID-19 decision-making.
Findings
Forecasts were used to address six types of policy questions, including epidemic spread, health system capacity, and intervention planning. Their perceived value depended on clarity, timeliness, and relevance. Preferences for metrics and formats varied: some valued forecasts for communicating uncertainty and exploring scenarios, especially in settings with strong modelling capacity. Others emphasized the role of expert briefings and locally adapted insights, especially where baseline data or technical capacity was limited. Across all settings, forecasts were most actionable when supporting binary decisions or scenario comparisons. Key barriers included poor data, late delivery, lack of contextual fit, and limited familiarity with modelling outputs.
Interpretation
These findings highlight the need for modular, user-facing tools that accommodate varying decision contexts and capacities. Embedding modellers in response teams, co-developing decision-relevant metrics, and strengthening health data systems are critical. Future pandemic responses will benefit from technical improvements and stronger institutional mechanisms for knowledge translation and collaboration.
Funding
This study was funded by Wellcome Trust and the UK MRC Centre for Global Infectious Disease Analysis, within the Global Health EDCTP3 Joint Undertaking.
RESEARCH IN CONTEXT
Evidence before this study
We searched PubMed and preprint databases (Jan 1, 2020–Sept 30, 2024) for publications on how epidemic forecasts were used, interpreted, or communicated in COVID-19 policy-making across settings, with no language restrictions. Existing literature shows that epidemic forecasts became important decision-support tools during COVID-19, but their uptake and influence on policy varied widely across countries and levels. Studies from high-income settings reported that use of models was often limited by lack of in-house modelling expertise, variable trust in model outputs, and poor alignment of forecasts with the practical questions policymakers faced. In low- and middle-income countries, evidence (including a systematic review in Africa) indicated that models did inform some decisions and were deemed critical to pandemic planning, yet technical capacity gaps and minimal engagement between modellers and policymakers constrained their usefulness. Overall, published analyses highlighted common needs to better integrate modelling with decision-making – for example, by improving collaboration and clearly communicating uncertainty – but few studies provided a global, cross-context perspective on these issues.
Added value of this study
To our knowledge, this is the first multinational study to systematically examine the policy use and interpretation of COVID-19 forecasts across diverse settings (46 countries, including high-income and resource-constrained contexts). Through an online survey (143 respondents) and follow-up interviews, we identified six common categories of policy questions that forecasts were used to address (such as projecting epidemic size, health system capacity, and intervention planning) and showed that a forecast’s perceived value depended heavily on its clarity, relevance, and timeliness. The mixed-methods approach captured a wide range of user perspectives, revealing widely varying preferences for metrics and presentation formats – for instance, some decision-makers valued forecasts for exploring uncertainty and scenarios (often in settings with strong modelling capacity), whereas others relied on expert briefings and locally tailored interpretations where data or modelling expertise were limited. Across all settings, we found that forecasts were most actionable when informing clear “yes/no” intervention decisions or comparing policy scenarios, and we pinpointed recurrent barriers to uptake (including poor data quality, late delivery of results, lack of contextual fit, and limited technical familiarity) that hampered effective use of forecasts, especially in low-resource environments. By capturing these insights, our study fills a gap in the literature and provides concrete recommendations to bridge the modeller–policy divide in future health emergencies.
Implications of all the available evidence
Taken together, prior studies and our findings underscore that improving the real-world impact of epidemic forecasts will require stronger collaboration, capacity, and communication between modelers and policymakers in both high-income and low-income settings. Bridging the “know–do” gap calls for institutional changes: for example, embedding forecasting teams within public health agencies or emergency operations, and co-developing models and metrics with end-users, so that projections directly address policy-relevant questions and constraints. Clearer communication of uncertainty and more user-friendly, context-tailored forecasting tools are also critical to ensure that decision-makers with varying technical backgrounds can interpret and trust the projections. Implementing these changes – investing in joint modelling-policy partnerships, strengthening data systems, and fostering knowledge translation mechanisms – will make epidemic forecasts more actionable and enhance evidence-informed decision-making for future outbreaks.
