Multi-level upstream prediction analysis of excessive drinking in the U.S. using machine learning and the American Nations cultural framework

In the United States, the health of a community is rarely determined by a single choice or a single factor. Instead, it emerges from a complex web where culture, economics, physical surroundings, and individual habits intertwine. Public health researchers have long understood that behaviors like drinking alcohol are not isolated acts but are deeply influenced by the environment in which people live. Some regions foster social habits that encourage gathering and celebration, while others face economic pressures that limit access to resources. When scientists try to understand why some places have high rates of excessive drinking and others do not, they must look beyond simple cause-and-effect. They need to see how dozens of different forces—from the quality of local food stores to the historical settlement patterns of a region—push and pull against each other to shape the daily lives of millions of people.

A new study brings a powerful set of digital tools to this challenge, aiming to untangle the specific mix of factors that drive excessive drinking across the United States. The researchers focused on the county level, analyzing data from nearly three thousand counties to build a detailed picture of what predicts heavy alcohol use. They used a type of advanced computer learning, known as machine learning, which allows a computer to find hidden patterns in massive amounts of data that traditional methods might miss. By feeding the computer information about health, economics, culture, and behavior, the team created a model that could explain seventy-two percent of the differences in drinking rates from one county to another. This high level of accuracy suggests that the model successfully captured the real-world complexity of the issue, offering a clearer view of the drivers behind a major public health problem.

The most surprising finding from this analysis concerns the relationship between chronic disease and drinking. The computer model identified the prevalence of diabetes as the strongest predictor of excessive drinking, but the connection was the opposite of what many might expect. In counties where more adults had diabetes, the rates of excessive drinking were actually lower. The researchers suggest this is likely because people diagnosed with a serious condition like diabetes often receive medical advice to change their habits, including cutting back on alcohol. Similarly, other factors often seen as signs of struggle, such as food insecurity, lack of physical activity, and insufficient sleep, were also linked to lower rates of heavy drinking. This points to a mechanism of economic constraint: when people are struggling to afford basic needs or are too exhausted to engage in social activities, they simply have less money and energy to spend on alcohol. These results challenge the common assumption that poverty and poor health always lead to higher rates of risky behaviors; in this case, they appear to act as a structural barrier to drinking.

Conversely, the study found that factors associated with social connection and better resources were linked to higher rates of excessive drinking. Counties with higher rates of smoking, more days of mental distress, and stronger social networks tended to have more heavy drinking. This aligns with the idea that alcohol and nicotine often reinforce each other in the brain's reward system, and that people may use alcohol to cope with psychological stress. Perhaps most notably, the study confirmed that the cultural identity of a region plays a massive role. Using a framework that divides the United States into distinct cultural nations based on their original settlers, the researchers found that areas known for strong community ties and social infrastructure, such as Yankeedom and the Midlands, had the highest rates of excessive drinking. This creates a paradox: these same regions often perform better on almost every other health metric, yet they struggle more with alcohol. The researchers propose that the very things that make these communities healthy—dense social networks, walkable neighborhoods, and accessible gathering places—also create more opportunities and social pressure to drink.

To ensure these patterns were not just coincidences, the team used a rigorous statistical method to test for cause and effect. They examined whether changing specific environmental factors would directly alter drinking rates. The analysis confirmed that increasing food insecurity or physical inactivity was associated with a decrease in excessive drinking, reinforcing the idea that economic hardship limits alcohol consumption. In contrast, improving the local food environment or increasing access to exercise opportunities was linked to a rise in drinking, likely because these improvements signal a more vibrant, socially active community where alcohol is more available and normalized. The study also found that smoking rates and mental health burdens were significant drivers, suggesting that interventions need to address these interconnected issues together rather than in isolation.

This research highlights that solving the problem of excessive drinking requires a nuanced understanding of local context. A one-size-fits-all approach will not work because the reasons people drink vary wildly depending on where they live and what their community values. In some places, the barrier to drinking is a lack of money or the presence of a chronic illness that demands lifestyle changes. In others, the driver is the very strength of the community itself, where social life and cultural norms make alcohol a central part of daily interaction. By using advanced computer models to map these complex relationships, public health officials can now design interventions that are tailored to the specific cultural and economic realities of each county. The goal is to move beyond broad assumptions and create strategies that respect the unique fabric of each community while addressing the specific factors that fuel excessive drinking in that specific place.