MENTAL HEALTH STATUS AND THIRD PLACES USE AMONG RURAL WORKING-AGE ADULTS IN THE UNITED STATES

Abstract

There is increasing urgency to address mental health needs, particularly through social determinants of health and modifiable lifestyle factors. Although third place use, going to public places outside of work or home, is a lifestyle behaviour that influences social health and feelings of belonging, little is known about if third place use is associated with mental health, particularly among rural working-aged adults who face increased risk for poor mental health outcomes, including suicide. This article presents data analysed from a representative sample of 1,135 rural adults ages 18 to 64 to determine whether mental health was associated with various levels of third place use. Results suggest that both going to third places and engaging with others are associated with better mental health among rural working-aged adults. Findings inform the growing body of literature on third places as social determinants of health and third place use as a modifiable lifestyle factor.

INTRODUCTION

Across rural and urban contexts there is growing urgency to better support mental health and well-being. Rural communities tend to have fewer local resources and services than their urban counterparts to address mental health needs, contributing to structural inequities in health (Chen et al., 2022, Weinzimmer et al., 2012). While individual- and group-based mental health services are essential components of comprehensive mental health care (Mutschler et al., 2022; Society of Clinical Psychology, 2022), there is increasing attention to apply frameworks that integrate macro-level, social determinants of health (SDOH) and consider modifiable lifestyle factors in developing population health approaches to promote mental health (Cohen, 2017; Kaplan et al., 2015). One emerging area of focus is on the role of third places – the physical places outside of the home and workplace where people gather, meet, and interact, such as libraries, parks, religious and faith-based organisations, or coffee shops – as one of the SDOH that influences well-being (Reed & Bohr, 2021). Interacting with others while using third places is a lifestyle behaviour that has been linked to both physical and social health, as well as overall quality of life (Jeffres et al., 2009; Oldenburg & Brissett, 1982). However, less attention has been given to whether a similar relationship exists for measures of mental health.

The majority of research on third places has focused on urban settings (Bagnall et al., 2023; Latham & Layton 2019) and older adult populations (Fong et al., 2021; Wiles et al., 2009). Moreover, most rural-focused studies were conducted in Europe (Iversen et al 2023; Mair 2009), Canada (Cabras & Mount 2017a; Cabras & Mount 2017b), and Australia (Bagnall et al., 2023; Heath & Freestone, 2022), without conceptualising third places as SDOH. The remaining gap in the literature leaves us with limited understanding of the potential role of third places as a macro-level SDOH for mental health and well-being among working-age adults living in rural areas in the United States (U.S.). Through the conceptual lens of social cohesion, we applied Berkman and colleague’s (2000) conceptual model of social networks and health (Berkman’s model) and collected data from a demographically representative sample of 1,135 rural adults ages 18 to 64 living in the U.S. to determine whether self-reported measures of mental health are associated with various levels of third place use. Evaluating the relationship between mental health and third place utilisation can help inform community-based, population level approaches to support mental health and improve health equity for rural working-age populations. In what follows, we first situate this study in the existing literature on rural mental health in the U.S., then discuss the role of third places for social and emotional health, and finally illustrate how concepts are theoretically integrated. This review of the research will show that community-based approaches to supporting mental health in rural areas is needed and that there is a growing body of literature exploring third places as a SDOH related to social connectedness, social isolation, and mental or emotional well-being.

LITERATURE REVIEW AND THEORY

Rural working-aged adults in the U.S. are at increased risk for adverse mental health outcomes and bear a greater burden associated with poor mental health than non-rural populations. While rates of mental illness are similar across rural and urban locales in the U.S., rural populations experience a greater symptom burden from mental distress, including higher rates of suicide (Randolph et al., 2023; Ehlman et al., 2022). In addition, addressing mental health needs within rural contexts can be particularly challenging. Access-related barriers include few and unequally distributed mental health providers (HRSA, 2024; Mauri et al., 2019; Randolph et al., 2023), long travel distances, lack of public transportation options, and private transportation limitations, such as unreliable vehicles, cost for gas/fuel, no driver’s licence, and time. (Syed et al., 2013; Wolfe et al., 2020). Moreover, rural residents pay higher out of pocket expenses when they do seek mental health services (Ziller et al., 2010). Even with increased access to mental health care via telehealth services, insufficient broadband access and household technology reduces accessibility (Vogels, 2021; Weinzimmer et al., 2021). Collectively, these factors severely restrict opportunities for delivering appropriate and timely mental health care in rural America. Although formal health services are important, social interactions play a significant role in mental health and well-being (Walton, 2014).

Third places, by their nature, foster social interactions and create opportunities for shared experiences that may otherwise not be available (Oldenburg, 1989; Oldenburg & Brissett, 1982). Oldenburg and Brissett (1982) argued that third places support social and emotional health as locations that foster a sense of belonging or attachment, describing them as places where people can be their authentic selves and feel more comfortable expressing their thoughts and emotions. Moreover, Oldenburg and Brissett (1982) hypothesised that through informal conversations, people express their feelings, process their experiences and responses to situations, thereby gaining a different perspective on the happenings in their lives. This aligns with the theoretical foundations of social networks, social support, and social cohesion, originating with Durkheim and integrated into Berkman’s model (2000), that links social interactions with health (House, 1987).

Research expanding on this early work emphasises the role of third places, as macro-level factors, in providing space for social connections, enabling recurrent interactions, facilitating bond formation, and generating supportive relationships (Finlay et al., 2019; Klinenberg, 2018; Walton, 2014). Third places can contribute to social and emotional health because they promote formation of both strong and weak social ties through which people can exchange social support (Thompson, 2018). In rural areas that have experienced the closure of large industries and subsequent population decline, third places can help prevent feelings of social isolation (Jeffres et al., 2009). Moreover, people can develop attachments to third places which contribute to their sense of identity, well-being, connection, and belongingness (Finlay et al., 2019; Hidalgo & Hernandez, 2001; Rosenbaum, 2006). The sense of attachment to place not only increases social cohesion but has been shown to fulfil emotional needs (Rosenbaum, 2006; Williams & Hipp, 2019). However, access to third places and the potential benefits from them are not equally distributed. Rural communities have substantially fewer third places than their urban counterparts (Rhubart et al., 2022).

Evidence reveals the mechanisms through which third places contribute to social and emotional well-being operate across socioecological levels, influencing more than just direct individual interactions. In some rural areas, the presence of facilities and meeting places have a symbolic function that translate into a broader sense of community well-being (Haski-Leventhal, 2009; Iversen et al., 2023). Moreover, one study found that knowing third places were available in local communities was important for overall quality of life, more so than the kind of third place itself (Jeffres et al., 2009). These previous works raise several points. First, while growing availability of contextual data sources are allowing researchers to advance ecological approaches to this work, such as availability of third places and individual health and well-being, limited data on actual third place use in surveys that also capture mental health is severely limited (Rhubart et al., 2023). Second, distinguishing the level of engaged use of third places is critical for determining whether third place use is a modifiable lifestyle factor that is associated with mental health. Finally, theoretical grounding is needed to explicate the pathways through which third places as SDOH influence mental health. However, prior studies have not specifically examined level of engagement in third places (Cabras & Mount, 2017b; Jeffres et al., 2009; Williams & Hipp, 2019) or used population-level conceptual frameworks, nor have standardised metrics been developed to measure or evaluate engaged third place use, such as the time spent and level of socialisation with others.

Theoretical framing

In this article, we adapt Berkman’s model (2000) to illustrate the relationship between third places, as macro-level SDOH, and individual-level health outcomes among rural working aged adults. This model emphasises the bidirectional pathways of influence between adjacent contextual levels in shaping health outcomes (Figure 1). It also illustrates how features of the macro-level environment have the potential to create structural inequities in health. In the original model (Berkman et al., 2000), however, the causal pathways through which third places influence mental health are not specified.

Figure 1: Conceptual model.

Note: Adapted from Berkman et al., 2000.

Social cohesion provides strong theoretical support to explain the relationship between third places and individual health outcomes. Social cohesion is a multidimensional construct that refers to the quality of social relationships and the presence of social bonds within a society or community (Schiefer & van der Noll, 2017). Numerous studies link social cohesion to health and well-being (Klein, 2013; McGowan, 2022; Williams & Hipp, 2019), including for rural populations (Bruhn, 2009). Characteristics of social cohesion include close social relations, emotional connectedness, a sense of belonging, and orientation toward the common good (Dragolov et al., 2013; Schiefer & van der Noll, 2017). Through the creation and maintenance of social cohesion, third places promote formation of social ties, social interaction and exchange of social support that shape health behaviours and, ultimately, health outcomes. Integrating social cohesion within Berkman’s model allows us to explore third places as macro-level SDOH that may contribute to structural inequities in mental health well-being.

While attention has been given to the relationship between the role of third places as SDOH and social or emotional health, very little research has examined if a similar association exists for measures of mental health. We could expect such a relationship to exist, given that previous work has shown separately that third place use is associated with greater social ties (Alidoust et al., 2019), that social ties can foster social cohesion and exchange of social support (Williams & Hipp, 2019), and that both social cohesion and support are associated with lower rates of depression and higher rates of community mental well-being (Echeverría et al., 2008; Yang et al., 2019). Research, however, fails to empirically evaluate the relationship between third place use and measures of mental health. Therefore, we examine if use of third places is a modifiable lifestyle behaviour that may be associated with mental health among rural populations. Specifically, in this article we determine whether self-reported measures of mental health are associated with various levels of third place use among rural working-age adults.

METHODS

Research design and data collection

For this quantitative analysis, we collected data from 1,135 working-age adults (ages 18–64) living in rural counties in the contiguous U.S. from September 2022 through December 2022 using a cross-sectional online survey with Qualtrics panels (Moss et al., 2023). Rurality was determined using rural-urban continuum codes (RUCCs) 4–9 (U.S. Department of Agriculture Economic Research Service [USDA ERS], 2025). RUCCs subdivide each U.S. state into political or administrative areas known as counties or county equivalents. Population, as well as economic and social integration with nearby urban centres (adjacency), are used to classify counties as either rural or urban. Urban counties (RUCCs 1–3) are those that fall within a large population nucleus, or major metropolitan area (e.g., New York, Los Angeles). All urban counties are classified according to metropolitan area population and not county population. Rural counties (RUCCs 4–9) are classified based on county population and adjacency to a metropolitan area (Table 1).

Table 1:

2013 rural-urban continuum codes

Code	Description
Urban counties
1	Counties in metro areas of 1 million population or more
2	Counties in metro areas of 250,000 to 1 million population
3	Counties in metro areas of fewer than 250,000 population
Rural counties
4	Urban population of 20,000 or more, adjacent to a metro area
5	Urban population of 20,000 or more, not adjacent to a metro area
6	Urban population of 5,000 to 19,999, adjacent to a metro area
7	Urban population of 5,000 to 19,999, not adjacent to a metro area
8	Urban population of fewer than 5,000, adjacent to a metro area
9	Urban population of fewer than 5,000, not adjacent to a metro area

Source: U.S. Department of Agriculture Economic Research Service, 2025.

We used quotas based on 2020 American Community Survey (ACS) estimates (American Community Survey [ACS], 2021) and 2013 ERS RUCCs (USDA ERS, 2025) to ensure the sample was representative of the rural U.S. working-age population. A poststratification weight was constructed to adjust for sample differences in age, sex, race, ethnicity, education, and income using the 2021 ACS estimates because the final sample had disproportionate shares of respondents with low income and low educational attainment. We report quality responses, rather than response rates, because invitations to participate that were not received or viewed cannot be accounted for using panels (Callegaro et al., 2014). Quality responses reflect completed surveys that were retained after meeting data quality thresholds. Surveys responses were evaluated for straight lining (i.e., same answers on matrix questions), speeding (took less than half the median time to complete the survey), inappropriate or spurious entries to text questions, and contradictory or mismatched responses. Of the 3,368 respondents who entered the survey and were deemed eligible to participate, 1,135 respondents completed the survey and were retained, representing a 33.7% quality completion rate. The research protocol and materials were reviewed and approved by the Pennsylvania State University Institutional Review Board. Research procedures were performed in compliance with all relevant laws and institutional guidelines. Informed consent was obtained from all research participants and precautions taken to protect the privacy and confidentiality of participants and their responses.

Measures

Third place use was measured by first asking participants, in the last month, how much time in an average week they spent at specific types of third places, excluding times they were there as a paid employee (e.g., restaurants, bars, religious or spiritual organisations, libraries, parks, etc.) (Table 2).

Table 2:

Types of third places

Third places measured
Religious and spiritual organisations Civic organisations Libraries Community centres or senior centres Parks/lakes Local schools Coffee shops, diners, cafes Fast food restaurants Dine-in restaurants Hair/nail salons or barbershops Bowling alleys Fitness/recreation facilities

Response options included: a) I don’t go to a place like this; b) 1–10 minutes; c) 11–30 minutes; and d) more than 30 minutes. For each third place that respondents visited, they were subsequently asked: “In an average week, how much time do you spend talking to other people in each of the following places: (exclude times you are in these places because you are a paid employee of the place)”. Response options included: a) Don’t talk with other people; b) 1–10 minutes; c) 11–30 minutes; and d) more than 30 minutes. A two-step process was used to develop the items for measuring third place use. Initial cut points to assess use of third places were informed by prior research on meaningful social interactions (Lou et al., 2022; Reis & Wheeler, 1991). Categories of third place use were then revised following internal pilot testing to ensure they reflected peoples’ everyday experiences of going to various third places.

For analyses, third place utilisation was operationalised by constructing a categorical variable that integrated both time spent in third places and time spent talking with others across all types of third places. This new variable of third place utilisation in an average week included the following categories: a) Goes and talks with others for more than 30 minutes in at least one third place; b) Goes and talks with others for 11–30 minutes in at least one third place; c) Goes and talks with others for 1–10 minutes in at least one third place; d) Goes to at least one third place, but doesn’t talk with others; and e) Does not go to any third places (Figure 2).

Figure 2: Construction of five-level categorical variable for third place usage.

Mental health status was the primary outcome of interest and measured using three self-reported items: a) the Patient Health Questionnaire for Depression and Anxiety (PHQ-4) (Kroenke et al., 2009); b) frequent mental distress (the number of days respondents reported their mental health was not good) (Cree et al., 2020); and c) self-rated mental health (SRMH) (Statistics Canada, 2021). The PHQ-4 is a validated, four-question instrument to screen for depression and anxiety (Kroenke et al., 2009). Participants reported how often they were bothered by the following symptoms during the prior two weeks: a) little interest or pleasure in doing things; b) feeling down, depressed, or hopeless; c) feeling nervous, anxious, or on edge; and d) not being able to stop or control worrying. Response options for each prompt included: 0) not at all, 1) several days, 2) more than half the days, and 3) nearly every day (Kroenke et al., 2009). Scores for depression or anxiety were calculated by summing the four numerical responses. Scores of six or more were recoded as screening positive for psychological distress, which is consistent with other approaches (Kroenke et al., 2009; Centers for Disease Control and Prevention [CDC], 2022). Frequent mental distress was assessed by asking participants, “Now thinking about your mental health, which includes stress, depression, and problems with emotions, for how many days during the past 30 days was your mental health not good?” (CDC, 2022). Participants who reported 14 days or more were coded as 1=Frequent mental distress, while those who reported 0–13 days were coded as 0=No frequent mental distress (CDC, 2022). Finally, SRMH was assessed by the question, “In general, would you say your mental health is?” (Statistics Canada, 2021). Responses were recoded as 1=Fair or poor SRMH and 0=Good, very good, or excellent SRMH (CDC, 2022). Psychometric testing has shown that SRMH is a valid and reliable metric to assess population level mental health symptom burden (Ahmad, et al., 2014).

Model covariates included age, sex, race and ethnicity, relationship status, presence of children in the home, educational attainment, employment status, income, and COVID hesitancy. For respondent sex, a “non-binary” option was included. With few participants selecting “non-binary”, sex was recorded as male vs. not male, with not male including those who selected both female and non-binary (Urlacher, 2023). Due to small numbers in some race and ethnicity categories, those identifying as non-Hispanic American Indian or Alaska Native, non-Hispanic Asian or Pacific Islander, or non-Hispanic other race(s) were combined into a non-Hispanic other race category. Education categories included high school degree or less, some college, and a bachelor’s degree or higher. Participants who reported that they had a job, whether part-time or full-time, were identified as employed and the remaining were recoded as not employed. Respondents who reported that they were married or a member of an unmarried couple were recoded as being part of a couple, with the remaining coded as single. Participants who reported having one or more children under the age of 18 living in the home were recoded as having children present. Household income was the total income from all sources before taxes and deductions and was collapsed into four categories: a) low income (less than $25,000); b) middle income ($25,000 to $49,999); c) and high income ($50,000 or more); and d) those who did not know their income or chose not to answer. Finally, we controlled for COVID-19 hesitancy given that persistent concerns about contracting illness in public spaces may have reduced the likelihood of using third places. To do this we used participant responses to the question: “Think about the last month. In an average week, did you do any of the following because of the COVID-19 pandemic: Chose to stay home instead of going to a public place.”

Data analysis

We summarised the unweighted and weighted distributions of the sociodemographic characteristics of the sample, model covariates, and mental health measures using descriptive statistics. Table 3 shows distributions of third place utilisation by level.

Table 3:

Weighted percentage distribution of third place utilisation (five-level)

	Doesn’t go	Goes, doesn’t talk	Goes, talks 1–10 minutes	Goes, talks 11–30 minutes	Goes, talks >30 minutes
All Third Places	9.42	14.19	24.59	18.47	33.34
Religious and spiritual organisations	63.63	5.33	13.50	8.35	9.18
Civic organisations	90.37	2.83	2.70	2.44	1.67
Libraries	73.89	7.14	13.20	4.54	1.23
Community centres or senior centres	90.00	2.58	3.03	3.04	1.35
Parks/lakes	48.47	18.72	17.06	7.50	8.25
Local schools	79.67	2.32	8.24	3.66	6.11
Coffee shops, diners, cafes	61.41	13.59	18.62	4.92	1.46
Fast food restaurants	23.84	32.87	36.58	5.11	1.6
Dine-in restaurants	40.08	18.46	23.31	11.90	6.26
Hair/nail salons or barbershops	69.68	5.87	10.25	7.42	6.79
Bowling alleys	84.52	4.27	5.37	3.07	2.76
Fitness/recreation facilities	77.21	5.12	7.68	4.52	5.47

Note: N=1,112

To determine if self-reported measures of mental health are associated with third place use among rural working-age adults, we constructed logistic regression models that predict the odds of reporting negative mental health status - for each of the three mental health metrics - using the five-category measure of third place use. Table 4 reports the unadjusted regression models.

Table 4:

Unadjusted logistic regression models predicting self-reported mental health by third place utilisation

	Model 1: Moderate, Severe Anxiety or Depression			Model 2: Frequent Mental Distress (14+ days)			Model 3: Poor, Fair Mental Health
	Odds Ratio	95% CI	p-value	Odds Ratio	95% CI	p-value	Odds Ratio	95% CI	p-value
Third Place Use (Ref: Doesn’t go)
Goes, don’t talk	0.648	(0.387, 1.085)	0.987	0.963	(0.586, 1.583)	0.882	0.624	(0.375, 1.037)	0.069
Goes, talks 1–10 minutes	0.578	(0.362, 0.925)	0.022	0.629	(0.397, 0.996)	0.048	0.439	(0.275, 0.699)	0.001
Goes, talks 11–30 minutes	0.440	(0.265, 0.730)	0.002	0.367	(0.222, 0.607)	<.001	0.250	(0.152, 0.411)	<.001
Goes, talks >30 minutes	0.434	(0.274, 0.686)	<.001	0.394	(0.250, 0.619)	<.001	0.215	(0.136, 0.341)	<.001
Sex (Ref: Female, non-binary)
Male	0.645	(0.493, 0.844)	0.001	0.623	(0.480, 0.809)	<.001	0.618	(0.481, 0.793)	<.001
Age	0.962	(0.952 ,0.972)	<.001	0.959	(0.949, 0.969)	<.001	0.955	(0.945, 0.965)	<.001
Race and Ethnicity (Ref: Non-Hispanic white)
Hispanic	1.530	(0.986, 2.375)	0.058	0.987	(0.626, 1.557)	0.955	1.490	(0.969, 2.293)	0.069
Non-Hispanic Black	0.816	(0.484, 1.376)	0.446	0.651	(0.386, 1.099)	0.108	0.692	(0.425, 1.127)	0.139
Non-Hispanic other	0.940	(0.519, 1.703)	0.839	1.396	(0.809, 2.409)	0.230	1.430	(0.831, 2.459)	0.197
Relationship Status (Ref: Separated, divorced, widowed)
Married or committed couple	0.678	(0.519, 0.887)	0.005	0.497	(0.383, 0.647)	<.001	0.471	(0.365, 0.606)	<.001
Children <18 in home (Ref: No)
Yes	1.091	(0.833, 1.430)	0.527	0.928	(0.713, 1.209)	0.581	1.031	(0.801, 1.328)	0.811
Educational Attainment (Ref: High school or less)
Some college	0.715	(0.527, 0.972)	0.032	6.060	(0.449, 0.818)	0.001	0.595	(0.446, 0.792)	<.001
Bachelor's degree or higher	0.494	(0.336, 0.727)	<.001	0.443	(0.305, 0.645)	<.001	0.460	(0.325, 0.625)	<.001
Annual Household Income (Ref: High Income=$50,000+)
Mid income ($25-$49,999)	1.676	(1.203, 2.336)	0.002	1.907	(1.383, 2.629)	<.001	1.863	(1.365, 2.541)	<.001
Low income (<$25,000)	1.852	(1.317, 2.604)	<.001	1.842	(1.320, 2.571)	<.001	1.981	(1.433, 2.737)	<.001
Don’t know income	1.074	(0.509, 2.265)	0.852	0.925	(0.440, 1.946)	0.837	2.335	(1.195, 4.563)	0.013
Employment Status (Ref: Not Employed)
Employed	0.688	(0.524, 0.904)	0.007	0.553	(0.424, 0.721)	<.001	0.633	(0.491, 0.817)	<.001
COVID Hesitancy (Ref: Went to public places rather than staying home)
Stayed home	1.432	(1.093, 1.877)	0.009	1.106	(0.848, 1.441)	0.456	1.134	(0.879, 1.464)	0.332

N=1,112

Sociodemographic covariates related to third place utilisation or mental health outcomes were identified from the literature and initially included in the models (Alegria et al., 2018; Carod-Artal, 2017; Walton, 2014). Best subset selection was then used to construct parsimonious models with the best fit (Hafermann et al., 2021; King, 2003). All analyses used a post-stratification weight to ensure that the overall sample was representative of the broader rural population in the U.S. by age, sex, race and ethnicity, education, and income.

Cases for which respondents reported more time talking with others than time spent in a third place (n=185) were recoded so that time spent socialising did not exceed, but rather matched, the time spent in each third place. For cases with missing values on the PHQ-4, missing values were converted to zero if doing so would not alter the composite score (n=1). All cases where the composite score for the PHQ-4 would be affected were excluded from analyses, as well as all cases with missing values for frequent mental distress or for SRMH. In total, 23 cases (2% of the sample) were excluded for an analytic sample size of 1,112. Statistical analyses were conducted using SAS software, version 9.4 of the SAS system for Windows (© 2020 SAS Institute Inc).

FINDINGS

Table 5 presents the unweighted and weighted distributions of the sample characteristics and the model covariates, as well as the ACS 2021 five-year estimates for the rural U.S. population, defined as RUCCs 4–9 (ACS, 2021), for selected measures For measures of sex, age group (18–34 years and 35–64 years), race and ethnicity, education, and income level, the final weighted sample approximated the rural U.S. population based on the 2021 ACS five-year estimates (ACS, 2021). Across other covariates, over half (57.31%) of respondents were either married or part of a committed couple, and less than half (41.50%) reported having children under age 18 residing in their home. A greater proportion of respondents were employed (58.41%) than not employed (41.59%). Nearly forty percent (39.01%) reported COVID hesitancy, indicating that they stayed home rather than going to a public place in the last month.

Table 5:

Descriptive Statistics of Sample Characteristics and Model Covariates

	Unweighted Frequency	Unweighted Percent	Weighted Percent	ACS 5-Year Estimates (2021) *
Demographic Characteristics
Sex
Male	524	47.12	50.50	51.24
Female or non-binary	588	52.88	49.50	48.76
Age group^
18–34 years	382	34.35	35.04	35.11
35–64 years	730	65.65	64.96	64.89
35–44 years	263	23.65	22.38	-
45–54 years	216	19.43	20.03	-
55–64 years	251	22.57	22.55	-
Race and Ethnicity
Non-Hispanic white	894	80.40	77.45	77.71
Hispanic	104	9.35	9.08	8.94
Non-Hispanic Black	73	6.56	7.95	7.92
Non-Hispanic other race(s)	41	3.69	5.52	5.43
Relationship status
Married or committed couple	567	50.99	57.31	-
Divorced or separated	172	15.47	13.09	-
Widowed	38	3.42	2.81	-
Single	335	30.13	26.79	-
Presence of children under age 18 in home
Yes	432	38.85	41.50	-
No	680	61.15	58.50	-
Socioeconomic Characteristics
Highest education
High school or less	537	48.29	47.26	48.11
Some college	390	35.07	31.29	30.90
Bachelor's degree or higher	185	16.64	21.45	21.00
Bachelor’s degree	142	12.77	13.77	-
Graduate degree or higher	43	3.87	7.71	-
Annual household income
Low income (<$25,000)	372	33.45	21.28	22.70
Mid income ($25,000-$49,999)	344	30.94	23.13	24.25
High income ($50,000+)	350	31.47	51.94	53.05
Don’t know income	46	4.14	3.64
Employment status
Employed	533	52.07	58.41	-
Not employed	579	47.93	41.59	-
COVID hesitancy: Going to public places
Stayed home rather than going in last month	431	38.76	39.01	-
Went rather than stay home in last month	681	61.24	60.99	-

Notes: N=1,112;

^*Residents of rural counties only;

^{^}Age was treated as a continuous variable in all analyses, the 18–34 and 35–64 categories were used for establishing sampling quotas and creating a sampling weight.

Unweighted and weighted distributions of self-reported measures of mental health status are reported in Table 6.

Table 6:

Self-reported mental health status

	Unweighted Frequency	Unweighted Percent	Weighted Percent
Anxiety/Depression Screen
None/Mild	760	68.35	72.59
Moderate/Severe	352	31.66	27.41
Poor Mental Health Days
<14 Days	727	65.37	68.82
14+ Days (frequent mental distress)	385	34.62	31.18
Self-rated Mental Health
Excellent/Very Good/Good	810	72.85	67.97
Fair/Poor	302	27.16	32.03

N=1,112

Across measures of mental health status, approximately one quarter (27.41%) had PHQ-4 scores that were moderate to severe, 31.18% reported frequent mental distress (14 days or more), and 32.03% reported fair or poor SRMH.

The fully adjusted logistic regression models predicting the odds of respondents reporting negative mental health states - anxiety/depression (psychological distress) (Model 1), frequent mental distress (Model 2), and fair/poor SRMH (Model 3) - by level of utilisation of all third places in an average week are presented in Table 7.

Table 7:

Adjusted logistic regression models predicting self-reported mental health status by third place utilisation

	Model 1: Moderate/Severe Anxiety or Depression			Model 2: Frequent Mental Distress (14+ days)			Model 3: Poor, Fair Mental Health
	Odds Ratio	95% CI	p-value	Odds Ratio	95% CI	p-value	Odds Ratio	95% CI	p-value
Third Place Use (Ref: Doesn’t go)
Goes, don’t talk	0.810	(0.469, 1.398)	0.449	1.227	(0.715, 2.106)	0.458	0.788	(0.455, 1.366)	0.396
Goes, talks 1–10 minutes	0.731	(0.440, 1.214)	0.226	0.772	(0.465, 1.282)	0.317	0.560	(0.336, 0.933)	0.026
Goes, talks 11–30 minutes	0.649	(0.375, 1.122)	0.122	0.533	(0.307, 0.928)	0.026	0.363	(0.211, 0.626)	0.000
Goes, talks >30 minutes	0.596	(0.360, 0.985)	0.044	0.541	(0.326, 0.897)	0.017	0.280	(0.168, 0.467)	<.0001
Sex (Ref: Female, non-binary)
Male	0.767	(0.575, 1.023)	0.071	0.730	(0.549, 0.970)	0.030	0.730	(0.555, 0.960)	0.024
Age	0.958	(0.947, 0.970)	<.0001	0.953	(0.941, 0.964)	<.0001	0.953	(0.942, 0.964)	<.0001
Race and Ethnicity (Ref: Non-Hispanic white)
Hispanic	1.212	(0.756, 1.943)	0.425	0.782	(0.478, 1.281)	0.329	1.093	(0.684, 1.747)	0.709
Non-Hispanic Black	0.630	(0.361, 1.099)	0.104	0.474	(0.267, 0.842)	0.011	0.486	(0.283, 0.833)	0.009
Non-Hispanic other	0.685	(0.360, 1.306)	0.251	0.877	(0.472, 1.627)	0.677	0.930	(0.504, 1.719)	0.818
Relationship Status (Ref: Separated, divorced, widowed)
Married or committed couple	0.862	(0.637, 1.167)	0.336	0.611	(0.453, 0.823)	0.001	0.589	(0.442, 0.786)	0.000
Children <18 in home (Ref: No)
Yes	0.936	(0.689, 1.271)	0.336	0.883	(0.650, 1.199)	0.425	0.943	(0.703, 1.267)	0.699
Educational Attainment (Ref: High school or less)
Some college	0.887	(0.636, 1.237)	0.481	0.774	(0.556, 1.076)	0.127	0.775	(0.564, 1.064)	0.115
Bachelor’s degree or higher	0.787	(0.509, 1.216)	0.280	0.788	(0.514, 1.209)	0.275	0.840	(0.562, 1.256)	0.396
Annual Household Income (Ref: High Income=$50,000+)
Mid income ($25-$49,999)	1.488	(1.037, 2.135)	0.031	1.583	(1.109, 2.258)	0.011	1.568	(1.110, 2.215)	0.011
Low income (<$25,000)	1.452	(0.978, 2.155)	0.064	1.188	(0.803, 1.759)	0.388	1.375	(0.938, 2.016)	0.103
Don’t know income	0.553	(0.249, 1.229)	0.146	0.391	(0.174, 0.876)	0.023	1.124	(0.541, 2.335)	0.754
Employment Status (Ref: Not Employed)
Employed	0.637	(0.464, 0.876)	0.006	0.500	(0.364, 0.685)	<.0001	0.587	(0.433, 0.797)	0.001
COVID Hesitancy (Ref: Went to public places rather than staying home)
Stayed home	1.448	(1.087, 1.929)	0.011	1.156	(0.867, 1.541)	0.324	1.161	(0.880, 1.533)	0.292
Somers’ D	0.349			0.391			0.442
c	0.674			0.695			0.721
AIC	1224.379			1231.802			1303.262
Max R-Square	0.147			0.216			0.252

N=1.112

For all three measures of mental health status, higher utilisation of third places was significantly associated with lower odds of negative mental health states when compared to those who did not use third places. Compared to those who did not use third places, those who reported going and spending more than 30 minutes talking with others in third places in an average week had 40.4% lower odds of psychological distress, 45.9% lower odds of frequent mental distress, and 72.0% lower odds of reporting fair/poor SRMH. Even spending 11–30 minutes talking with others in a third place in an average week lowered the odds by 46.7% for frequent mental distress and 63.7% for poor or fair SRMH compared to those who did not use third places. Going to and spending 1–10 minutes talking with others at third places was only significant for reducing the odds for poor or fair SRMH (44.0%).

Discussion

Third places have the potential to foster supportive relationships that promote social and emotional health and may play a particularly important role in rural communities (Cabras & Mount, 2017a; Finlay et al., 2019; Klinenberg, 2018). Until now, we have had very little understanding of third places as macro-level SDOH or the relationship between use of third places and mental health status in rural America. Therefore, in this study we collected data from 1,135 working-age adults residing in rural counties of the contiguous U.S. to examine the relationship between self-reported measures of mental health and different levels of third place use using Berkman’s model (2000) and the conceptual lens of social cohesion. Results confirmed our hypothesis that average weekly use of third places that include engaging in conversation with other people for a substantial time (i.e., >30 minutes) was associated with better measures of mental health than those who do not use third places. The results suggest engaging with others in third paces may be important for mental health among rural working-age adults, highlighting the role of third places as SDOH (Bagnall et al., 2023) and suggesting their contribution to structural inequities in health outcomes for rural communities. This is important given higher rates of suicide and lower access to mental health services in rural contexts (Randolph et al., 2023). Results align with prior research tying social health and well-being to third place use (Klinenberg, 2018; Oldenburg & Brissett, 1982). Findings also provide a first step in suggesting that meaningful third place use – where conversations happen with other people – may be a modifiable lifestyle behaviour for supporting mental health.

It is important to note that while going and spending >30+ minutes talking with others in a third place was positively associated with all three mental health measures, the effect varied across middle categories of third place use. The odds of frequent mental distress and fair or poor SRMH were reduced among respondents socialising for 11–30 minutes, and only for SRMH for those talking with others for 1–10 minutes. This may suggest that different levels of third place use influence different domains of mental health and should be explored in future research. This may also reflect varying levels of social cohesion across individuals or groups.

Further, for this sample of rural working-aged adults, going to third places and not interacting with others was not associated with any of the self-reported measures of mental health. This finding does not fully align with prior research and the social cohesion literature showing indirect social and emotional benefits attributed to the presence or availability of third places in communities (Iversen et al., 2023; Jeffres et al., 2009). While we were unable to assess whether availability of third places played a role in the relationship between third place use and mental health, the overall findings point to greater mental health benefits with higher levels of engaged use of third places. This suggests a possible threshold for socialising and third places – meaningful use – at which social and emotional benefits are realised.

These findings have implications for policy and practice. The significance of third places as SDOH may inform community development efforts. Local leaders could consider working with local businesses and non-profits to preserve existing third places and create welcoming spaces or opportunities for people to connect. State-level leaders could also target block grants and other funding sources to communities with limited third place infrastructure and high rates of mental illness, especially within rural contexts. This may include supporting improved transportation options (e.g., public transportation, ride sharing, and ride-hailing businesses) or creating free or low-cost opportunities in existing public places (e.g., schools, parks, or libraries) to reduce structural inequities, particularly given the economic instability, third place loss, and population decline experienced by many rural U.S. communities (Brown & Schafft, 2019; Parsons, 2022). Although outside the scope of this article, such efforts can foster economic development, which may also bolster the attractiveness of rural spaces and reduce barriers for recruiting mental health professionals.

This research contributes to the growing body of research explaining the pathways through which factors of macro-, meso-, and micro-levels shape health outcomes (Klein, 2013). Specifically, this study addresses previously identified needs to identify contextual factors, like SDOH, as well as the pathways through which they operate to understand their influence on mental well-being (Cohen, 2017; Finlay et al., 2019; Kaplan et al., 2015), particularly those that may contribute to structural inequities in health for rural U.S. populations (Algeria et al., 2018; Berkman et al., 2000). The work expands existing theory, providing empirical support for the relationship between third places and mental health described in Berkman’s model (2000) and proposing social cohesion as the pathway through which third places operate to influence health outcomes in rural communities.

Although this research provides the first examination of various levels of third place use and self-reported measures of mental health status in rural America using a nationally representative sample, there are several limitations. First, assessing mental health status at the population level is challenging and there is concern that general measures of mental health are not ethno-racially appropriate (Assari & Lankarani, 2017; Santos-Lozada & Martinez, 2018). To address this, we controlled for race and ethnicity to limit potential effects of culturally based differences in perception or interpretation of the mental health questions. Second, recoding missing values to questions from the PHQ-4 and values for time spent talking with others that exceeded time spent in third places may have introduce error. Unreported analyses using the original and recoded values, however, showed no substantial effect on the results. Third, data are cross-sectional and self-reported. Therefore, it is possible that third place utilisation was under-reported based on respondents’ interpretations of third place categories. In addition, distance to third places and transportation availability are important contextual factors that can impact third place utilisation, yet remained outside the research scope. Finally, selection bias may be at play – those who have better mental health may also be more likely to go to and engage with others in conversation in third places. Future research should employ methods and measures to account for these limitations, including longitudinal and quasi-experimental designs.

CONCLUSIONS

Rising concerns about mental health have been met with calls to apply frameworks that integrate macro-level factors, like SDOH, and modifiable lifestyle factors in approaches to improve mental health equity. The use of third places – which has been shown to support feelings of connectedness, social health, and belonging – is one potential modifiable lifestyle behaviour that can be shaped through community efforts. This article provides the first quantitative evidence from a nationally representative sample of rural working-age adults in the United States to introduce a novel measure of third place utilisation and demonstrate an empirical relationship between engaged third place use and mental health. Findings highlight the potential importance of third places for mental well-being in rural communities and inform the growing body of literature on the vital role of third places as part of the built environment for rural populations, their influence on health outcomes, and possible driver of structural inequities.