Influenza can race through a crowded barracks or prison while circulating far more slowly in the wider population. Ebola may become substantially more difficult to contain when hospital beds fall below the levels routinely available in wealthier countries. Even within the same healthcare system, unequal treatment at the bedside can influence who survives. These are not separate from the biology of epidemics. They are part of the conditions that determine how pathogens move through communities and how severe their consequences become.
For decades, epidemiologists and social scientists have documented the connection between inequality and infectious disease. People’s risk of exposure can depend on where they live, how many people share their home, whether they can work remotely, and how easily they can reach a clinic. Access to vaccines, testing, treatment, transportation, and paid sick leave can further shape the course of an outbreak. At the same time, mathematical epidemiologists have built increasingly sophisticated models to forecast transmission and evaluate public-health interventions. Yet the social causes of unequal risk have often remained separate from the formal machinery used to model epidemics.
A new study in Biology Letters seeks to close that gap with a metric called structural causal influence, or SCI. The framework is designed for use with standard disease-transmission models and quantifies how strongly social determinants of health alter the behavior of an epidemic. Rather than treating inequality as background information, SCI allows researchers to represent factors such as crowded housing, limited vaccine access, financial constraints, and unequal healthcare availability as mechanisms that can directly change transmission and disease outcomes.
The approach is built around a central distinction in causal science: observing that two conditions occur together is not the same as showing that one helps produce the other. A disadvantaged community may experience higher infection rates because of multiple overlapping factors, including increased exposure, delayed access to care, or reduced ability to isolate. SCI is intended to help identify the influence of such structural conditions within a mathematical model. By comparing how the model behaves when a social factor is present, altered, or removed, researchers can estimate how that factor changes the spread of infection and the likely effect of an intervention.
The researchers tested the concept using canonical models such as SIR, which divides a population into susceptible, infected, and removed groups. In its simplest form, an SIR model describes how individuals move between these categories as infection spreads and recovery or removal occurs. The model’s basic parameters include the rate at which susceptible people encounter infectious individuals and the rate at which infected people recover or are otherwise removed from transmission. Although such models are deliberately streamlined, they can be extended to represent multiple communities, different contact patterns, unequal access to vaccination, and varying levels of healthcare capacity.
That added structure matters because a population-wide average can conceal the conditions in a smaller, high-risk group. Modeling with SCI showed that an epidemic may develop even when the overall transmission risk appears low. If infection spreads efficiently within a disadvantaged subgroup, that group can sustain transmission despite a reassuring average for the population as a whole. Depending on the connections between groups, infections may then spill into the wider community. A low average risk, in other words, does not necessarily mean that every population segment is protected from sustained outbreak dynamics.
This result has important implications for the interpretation of familiar epidemic indicators. Measures such as an average reproduction number summarize transmission across a population, but they can obscure differences in contact rates, susceptibility, vaccination coverage, or access to treatment. A single value may therefore suggest that an outbreak is under control while transmission remains intense in a particular neighborhood, institution, or occupational group. SCI offers a way to examine how those differences contribute to the overall dynamics and whether reducing them would change the trajectory of the epidemic.
The framework also provides a tool for evaluating public-health choices. An intervention that appears inefficient when judged only by its immediate effect on the population average may have a larger indirect benefit if it protects a group that is sustaining transmission. Increasing vaccine access in underserved communities, improving ventilation in crowded settings, expanding hospital capacity, or reducing financial barriers to testing and isolation could affect both direct health outcomes and the broader network of transmission. The metric is intended to help decision-makers compare these effects and determine where limited resources may produce the greatest reduction in disease spread.
The authors deliberately chose interpretable mathematical models rather than embedding the metric in complex artificial-intelligence systems. Neural networks can identify patterns in large datasets, but their internal reasoning may be difficult to inspect, making it harder to determine why a prediction was produced or which social conditions shaped it. By working with models that researchers and policymakers can read, SCI is designed to make assumptions visible and causal relationships easier to examine. The same principle could allow the framework to be adapted beyond acute epidemics, including studies of chronic disease, pollution, and climate-related health risks.
Brandon Ogbunu of the Santa Fe Institute and Yale University, the study’s senior author, says the findings reinforce the idea that infectious-disease control cannot be separated from equity. Sam Scarpino of Northeastern University, a co-author and Santa Fe Institute External Professor, describes the work as an example of how epidemiology, statistics, and data science can be combined to address questions that no single discipline can resolve alone. The researchers argue that directing more protection and healthcare resources toward communities facing the greatest structural disadvantages is not only a matter of fairness. Because outbreaks move through connected populations, reducing transmission in those communities can ultimately help protect everyone.
Subject of Research: Social inequality and its causal influence on infectious-disease transmission models
Article Title: Structural causal influence (SCI) captures the forces of social inequality in models of infectious disease
News Publication Date: 19 August 2026
Web References: https://santafe.edu/people/profile/ogbunu-brandon ; https://www.santafe.edu/people/profile/sam-scarpino
References: Biology Letters, “Structural causal influence (SCI) captures the forces of social inequality in models of infectious disease,” publication date: 19 August 2026
Image Credits: Edson De la O / Santa Fe Institute
Keywords: infectious disease, viral transmission, epidemic modeling, social inequality, health disparities, structural causal influence, SCI, SIR model, public health, vaccination access, healthcare access, epidemiology, computational biology
Tags: disparities in vaccination and testing accesseffects of overcrowding on disease transmissionhealth disparities in infectious disease spreadhealthcare access and epidemic containmentimpact of socioeconomic factors on disease outcomesinequality in healthcare resource distributioninfluence of living conditions on infectious disease riskintegrating social factors into epidemic modelsmodeling social determinants of health in epidemicssocial determinants of epidemic severitySocial inequality and epidemic transmissionstructural causal influence (SCI) in epidemiology
