Proposal Description

Randomized Controlled Trials (RCTs) are widely regarded as the gold standard for evaluating intervention effectiveness. However, they are often impractical, unethical, or prohibitively expensive for assessing large-scale public health policies, health system reforms, and population-level programmes. Increasingly, governments rely on data generated through national health surveys, population-based surveillance systems, administrative records, and Routine Health Information Systems (RHIS) to guide decision-making. This invited session will showcase modern statistical methods that generate credible causal evidence from such observational and routinely collected data, enabling more informed public health policy and resource allocation.

The session will bring together advances from causal inference, biostatistics, epidemiology, survey statistics, and official statistics to address the methodological challenges inherent in government-generated data. These challenges include selection bias, missing data, measurement error, time-varying confounding, and the limitations of non-randomised study designs. Presentations will demonstrate state-of-the-art approaches such as target trial emulation, marginal structural models, g-computation, difference-in-differences, regression discontinuity designs, synthetic controls, and propensity score methods. Particular attention will be given to integrating these approaches with complex survey designs and routine administrative data systems to strengthen causal interpretation and policy relevance.

The session is especially relevant to sub-Saharan Africa, where countries are investing heavily in digitized health information systems, demographic and health surveys, civil registration systems, and integrated data platforms such as DHIS2. These developments provide unprecedented opportunities to evaluate public health programmes and monitor progress toward national and global development goals. Yet substantial analytical capacity gaps remain in translating these data into robust evidence for decision-making. Through applied case studies, presenters will demonstrate how causal inference methods can be used to evaluate interventions and policies related to HIV, tuberculosis, malaria, maternal and child health, immunisation, nutrition, environmental health, and health systems strengthening.

By moving beyond simple correlations and descriptive analyses, these methods enable researchers and policymakers to estimate the true impact of policies and programmes, quantify counterfactual outcomes, and identify the most effective strategies under resource constraints. The session will therefore provide practical guidance on leveraging government surveys and routine data to support evidence-based governance and public health action.

The proposed session aligns closely with the objectives of the World Statistics Congress by demonstrating how modern statistical science can transform official statistics and administrative data into actionable evidence to improve population health and societal well-being. Bringing together statisticians, epidemiologists, official statisticians, and policymakers, the session will foster interdisciplinary dialogue and promote innovative applications of causal inference for the public good.