Understanding county-level mental health determinants to guide community prevention efforts

Abstract

Background

Mental health disparities represent a growing challenge for public health systems, making it important to understand how local environmental and socioeconomic conditions contribute to population mental health outcomes. This study examined how county-level environmental, socioeconomic, and demographic characteristics relate to mental health outcomes in the US.

Methods

Multivariable ordinary least squares regression was used with state-clustered standard errors. Two measures of mental health were analyzed: the average number of mentally unhealthy days and the percentage of adults experiencing frequent mental distress.

Results

Severe housing problems and insufficient sleep were consistently associated with poorer mental health, while higher food environment quality was linked to reduced frequent mental distress. Socioeconomic factors, including poverty, education, and median household income, showed clear associations. Demographic characteristics, such as higher proportions of non-Hispanic White and female residents, also corresponded to increased mental health challenges.

Conclusions

These findings identify county-level conditions that may serve as targets for community-based prevention and intervention strategies, providing guidance for program administrators, planners, and policymakers seeking to reduce mental distress and improve population mental health.

Mental health has become an increasingly important concern for policymakers, healthcare systems, and managers, as mental distress affects not only individual well-being but also labor productivity, healthcare utilization, and social stability.¹ In the United States, substantial variation in mental health outcomes exists across geographic areas, reflecting differences in living environments, socioeconomic resources, and population characteristics.^2,3 Counties, as key administrative and policy-relevant units, play a central role in shaping local conditions through public service provision, infrastructure investment, and coordination of health and social programs. Understanding how county-level conditions relate to mental health outcomes is therefore critical for designing effective interventions and allocating resources efficiently.

Mental health outcomes are influenced by a broad set of structural and contextual factors that extend beyond clinical care. Living environment conditions such as housing stability, access to nutritious food, and sleep adequacy shape daily stress exposure and coping capacity. Socioeconomic conditions—including poverty, education, income, and social support—affect access to resources, resilience to stress, and long-term mental well-being. At the same time, demographic composition, including age structure, gender, racial makeup, and rurality, may influence how mental distress is experienced and reported within communities. While prior research has documented associations between these factors and mental health, less attention has been paid to how different dimensions of mental health burden respond to the same set of county-level characteristics.

This study focused on two complementary measures of mental health: the average number of mentally unhealthy days and the percentage of adults experiencing frequent mental distress. Although closely related, these measures capture distinct aspects of population mental health. The average number of mentally unhealthy days reflects the overall intensity of mental health burden across the population, whereas frequent mental distress captures the prevalence of more persistent or recurrent mental health problems. Examining both outcomes jointly allows for a more nuanced assessment of whether county-level conditions influence general mental well-being and severe mental distress in similar or different ways.

Using nationally representative county-level data, this study examined how environmental and living conditions, socioeconomic factors, and demographic characteristics were associated with two key mental health outcomes. By estimating multivariable regression models with state-clustered standard errors, the analysis provides a comprehensive assessment of the structural determinants of mental health across US counties. The study’s goal is to highlight county-level conditions that may serve as actionable targets for community-based prevention and intervention efforts, offering guidance to program administrators, planners, and policymakers seeking to reduce mental distress and improve population mental health.

METHODS

We merged two publicly available datasets. The first dataset was the County Health Rankings National Data, obtained from the County Health Rankings & Roadmaps website (released January, 2025).⁴ The second dataset was the County-Level Poverty Data (released January 31, 2025), obtained from the US Department of Agriculture Economic Research Service.⁵ The County Health Rankings dataset provided county-level information for 3159 counties across the 50 US states, including two mental health outcome measures: the average number of mentally unhealthy days and the percentage of adults experiencing frequent mental distress, the latter of which is defined as the percentage of people reporting 14 or more days of poor mental health per month. In addition, this dataset included a comprehensive set of county-level covariates capturing living environment conditions (primary care physician ratio, mental health provider ratio, food environment index, percentage of severe housing problems, insufficient sleep, share of long-commute workers who drive alone, and average daily fine particulate matter [PM2.5]), social and economic factors (education, median household income, and lack of social and emotional support), and demographic characteristics (age 65 and over, female, non-Hispanic White, and percentage of people living in rural). The US Department of Agriculture dataset provided a state-level poverty measure, which was merged into the county-level data to account for broader socioeconomic conditions operating at the state level.

We first conducted descriptive analyses of both dependent variables to assess their overall distributions across counties. Histograms were used to display the frequency distributions of average mentally unhealthy days and the percentage of adults reporting frequent mental distress, providing an overview of central tendencies, dispersion, skewness, and potential outliers. These analyses characterized the extent of variation in mental health outcomes at the county level and informed the subsequent multivariable modeling.

To further examine bivariate relationships, we analyzed the distributions of key environmental, socioeconomic, and demographic covariates conditional on each mental health outcome. Visualizations were used to assess how county-level characteristics varied across different levels of mental health burden, offering an intuitive depiction of potential associations between explanatory variables and outcomes. This exploratory step provided preliminary insights into underlying patterns and helped identify relationships to be examined in the regression analysis.

Given the hierarchical structure of the data, with counties nested within states, we estimated multivariable ordinary least squares (OLS) regression models with standard errors clustered at the state level. This approach accounted for within-state correlation arising from shared policy environments, economic conditions, and institutional factors. Separate regression models were estimated for each dependent variable to enable direct comparison of coefficient magnitudes and statistical significance across outcomes.

Together, the descriptive analyses and multivariable regression models provide complementary perspectives on the determinants of mental health, allowing us to assess whether and how environmental, socioeconomic, and demographic factors differentially relate to the intensity versus prevalence of mental distress. All analyses were conducted using Python 3.12.12 in Google Colab.

RESULTS

For the distribution of average mentally unhealthy days, most counties clustered between 5 and 6 days, suggesting that this interval represented the modal level of mental health burden nationwide. A comparable pattern emerged for the prevalence measure, with most counties reporting frequent mental distress rates between 18% and 19%. Taken together, these findings indicate that mental health burden is relatively homogeneous across counties, with most experiencing moderate levels of mental distress, reflecting a broadly consistent pattern of mental health challenges at the county level (Figure 1).

Figure 1.

Distributions of mental health outcome variables across counties.

To explore bivariate relationships, we examined the distributions of key covariates stratified by each mental health outcome, providing insight into how county-level characteristics vary across different levels of mental health burden. Figure 2 presents the distributions of county-level covariates by the average number of mentally unhealthy days. Several factors—including rural residence and lack of social and emotional support—exhibited positive associations with the average number of mentally unhealthy days, whereas other covariates, such as the food environment index, education, and median household income, displayed negative associations.

Figure 2.

Distributions of county-level covariates across average number of mentally unhealthy days.

Figure 3 shows the corresponding distributions of county-level covariates across the percentage of adults experiencing frequent mental distress. The overall patterns closely mirrored those observed for the average number of physically unhealthy days. This similarity was expected, given the strong correlation between the two outcome measures, both of which are derived from the same underlying self-reported mental health construct.

Figure 3.

Distributions of county-level covariates across percentage of people experiencing frequent mental distribution.

Despite these similarities, the two measures capture distinct dimensions of mental health burden. The average number of mentally unhealthy days reflects overall population health by summarizing the mean number of poor mental health days, whereas the percentage of adults experiencing frequent mental distress focuses on individuals with more persistent or severe health problems, defined as reporting 14 or more unhealthy days within a month. Consequently, while the bivariate relationships were broadly consistent across outcomes, certain covariates differed in the strength and statistical significance of their associations between the two measures.

To assess the relationships between county-level characteristics and mental health outcomes, we fitted multivariable OLS regression models with standard errors clustered at the state level. Focusing first on the average number of mentally unhealthy days, the results revealed several covariates with statistically significant positive associations. Severe housing problems were associated with an increase in mentally unhealthy days, with a coefficient of 0.021 (P = 0.02). Lacking social and emotional support was associated with an increase in mentally unhealthy days, with a coefficient of 0.044 (P = 0.02). Insufficient sleep also showed a positive association (coefficient = 0.031, P = 0.01). In addition, counties with a higher proportion of non-Hispanic White residents reported more mentally unhealthy days (coefficient = 0.014, P < 0.001), as did counties with higher poverty rates (coefficient = 0.080, P = 0.02) and females (coefficient = 0.018, P = 0.02).

Conversely, a number of factors were negatively associated with the average number of mentally unhealthy days. Higher educational attainment was linked to fewer unhealthy days (coefficient = −0.012, P < 0.001). Median household income was similarly negatively related to mentally unhealthy days, although the magnitude was small (coefficient = −0.000003, P = 0.04).

Results differed when examining the percentage of adults experiencing frequent mental distress. Several county-level characteristics were positively associated with this outcome. Severe housing problems were strongly related to higher levels of frequent mental distress (coefficient = 0.057, P = 0.02), as was insufficient sleep (coefficient = 0.189, P < 0.001). Counties with higher non-Hispanic White populations reported a higher percentage of adults experiencing frequent mental distress (coefficient = 0.042, P < 0.001), as did counties with a higher proportion of female residents (coefficient = 0.103, P < 0.001) and higher poverty rates (coefficient = 0.184, P = 0.04).

In contrast, education, food environment, and median household income were inversely related to frequent mental distress. Higher educational attainment was associated with a lower percentage of adults experiencing frequent mental distress (coefficient = −0.055, P < 0.001). A higher food environment index was similarly protective (coefficient = −0.263, P = 0.005). A higher median household income was also negatively associated with this outcome (coefficient = −0.00001, P = 0.006). All regression results are summarized in Table 1.

Table 1.

Ordinary least squares regression results for mental health outcomes

Covariates	Average number of mentally unhealthy days		% frequent mental distress
	Coefficients	P value	Coefficients	P value
Primary care physicians ratio	−0.000007	0.12	−0.0000008	0.95
Mental health provider ratio	−0.00001	0.47	−0.00004	0.27
Food environment index	−0.052	0.07	−0.263	0.005
Insufficient sleep	0.031	0.01	0.189	<0.001
% Severe housing problems	0.021	0.02	0.057	0.02
% Long commute by driving alone	0.002	0.24	0.001	0.79
Average daily fine particulate matter	0.0004	0.98	0.008	0.88
% Poverty	0.080	0.02	0.184	0.04
% Education	−0.012	<0.001	−0.055	<0.001
Median household income	−0.000003	0.04	−0.00001	0.006
% Lacking of social and emotional support	0.044	0.02	0.849	0.12
% ≥65 years	0.004	0.94	−0.015	0.35
% Female	0.018	0.02	0.103	<0.001
% Non-Hispanic White	0.014	<0.001	0.042	<0.001
% Rural	−0.001	0.13	0.004	0.20

To ensure valid inference, we estimated OLS regressions with standard errors clustered at the state level, accounting for both heteroskedasticity and within-state correlation. Across specifications, the models exhibited strong explanatory power, with R² values of 0.612 and 0.717 (N = 2145), and were jointly significant (F-statistics = 104.6 and 131.1, respectively; P < 0.001 in both cases). The increase in explanatory power in the expanded model suggests improved fit after incorporating additional covariates, while maintaining overall model stability. Multicollinearity was assessed using variance inflation factors, with all explanatory variables consistently below 5 and mostly below 3, indicating no evidence of severe multicollinearity. These diagnostic results suggest that the model provides reliable statistical inference.

DISCUSSION

This study provides evidence that county-level characteristics are systematically associated with mental health outcomes across US counties. In particular, environmental and socioeconomic factors—including housing-related conditions, sleep adequacy, and economic resources—show more consistent and robust associations with both mentally unhealthy days and frequent mental distress than demographic characteristics. Although the two mental health measures capture related but distinct dimensions of distress, the results indicate that some county-level factors exert broad influences across both outcomes, whereas others are more outcome-specific. Taken together, these findings suggest that population mental health disparities are closely linked to modifiable structural conditions at the community level.

Environmental and living conditions show clear and differentiated associations with county-level mental health outcomes. Several environmental factors are consistently and positively associated with both the average number of mentally unhealthy days and the percentage of adults experiencing frequent mental distress. In particular, severe housing problems and insufficient sleep emerge as robust predictors across both outcomes, suggesting that housing instability and sleep deprivation represent fundamental stressors that affect both general mental well-being and more severe forms of mental distress.^6–8 These conditions likely operate through pathways of chronic stress, uncertainty, and physiological strain, resulting in broad population-level effects. In contrast, food environment quality is negatively associated with frequent mental distress but not with the average number of mentally unhealthy days. This pattern suggests that access to healthier food options may be more relevant for preventing persistent or severe mental distress rather than reducing overall distress levels in the general population.^9,10

Socioeconomic conditions are also closely linked to mental health outcomes, with several factors exhibiting consistent associations across both measures. Poverty is positively associated with both mentally unhealthy days and frequent mental distress, highlighting the mental health burden imposed by economic hardship and chronic financial stress.^11,12 Higher educational attainment is negatively associated with both outcomes, indicating a protective effect across different levels of mental health severity, potentially through improved health literacy, coping strategies, and access to resources.^13,14 Median household income likewise shows a negative association with both outcomes, although the estimated coefficients are small in magnitude due to the scale of the income measure.¹⁵ Lacking social and emotional support is significantly associated with mentally unhealthy days but not with frequent mental distress,^16,17 suggesting that deficits in social support may contribute to widespread emotional strain without necessarily leading to persistent or severe mental health problems. Demographic characteristics display more heterogeneous patterns. Counties with higher proportions of non-Hispanic White residents and female residents report greater mental health burden across both outcomes, which may reflect contextual or regional factors, differences in mental health awareness, or reporting behaviors rather than directly modifiable risk factors.^18,19

Several limitations should be noted. The cross-sectional design precludes causal inference, and all associations should be interpreted as correlational. For example, for factors such as insufficient sleep, it is difficult to determine whether they function as an etiological driver or a symptomatic manifestation of mental distress, as most mental health conditions are inherently associated with sleep disturbances. Second, the use of county-level aggregated data may mask within-county heterogeneity and raises the possibility of ecological fallacy. Specifically, an ideal analysis would utilize participant-level data within a three-level hierarchical model (nesting individuals within counties and states). However, such granular information is currently unavailable in these national surveillance datasets. In addition, mental health outcomes are self-reported and may be subject to reporting bias.

This study provides evidence that county-level environmental and socioeconomic conditions are more consistently associated with population mental health outcomes than demographic characteristics. The findings suggest that mental health disparities are largely driven by modifiable structural factors, particularly housing stability, sleep adequacy, food environments, and economic resources. From a community and policy perspective, interventions targeting these structural conditions may offer the greatest potential to reduce mental distress and improve population mental health across counties. For instance, enhancing local housing quality standards and expanding affordable housing initiatives could help mitigate the mental health burdens associated with severe housing problems. Addressing sleep inadequacy might involve community-wide education on sleep hygiene or urban planning that reduces environmental noise in residential areas. To combat the effects of food insecurity, local policies could focus on increasing the availability and affordability of nutritious options in underserved tracts.

In conclusion, this study examined county-level mental health outcomes using two complementary measures—the average number of mentally unhealthy days and the percentage of adults experiencing frequent mental distress—to capture both the intensity and prevalence of mental health burden. The findings showed that environmental and living conditions as well as socioeconomic factors, including housing stability, sleep adequacy, food access, poverty, education, income, and social support, are consistently associated with poorer mental health outcomes across counties. In contrast, demographic characteristics display more heterogeneous and outcome-specific associations, suggesting that demographic composition shapes the distribution of mental distress rather than serving as a primary driver of overall mental health burden. From a management and policy perspective, these results indicate that county-level mental health disparities are more strongly influenced by modifiable structural and socioeconomic conditions than by demographic factors alone. Policies and managerial interventions that improve living environments and socioeconomic resources may therefore offer more effective and scalable pathways for enhancing population mental health.

Disclosure statement/Funding

The authors report no funding or conflict of interest.