Abstract
Long-term residence in neighborhoods is thought to promote the development and maintenance of supportive relationships and trust. These strong social ties may, however, be limited in communities in post-industrial cities characterized by high levels of vacant properties. This study aimed to examine the relationship between neighborhood vacancy and mental health with adjustment for length of residence and possible moderation by social (dis)integration in a sample of Flint, MI, residents. We found that short-term (but not long-term) increases in neighborhood vacancy were associated with poorer mental health, after adjustment for individual covariates. When considering neighborhood vacancy, length of residence and individual covariates, however, the only significant association detected was between higher social disintegration and lower wellbeing. This effect was direct and not mediated by other factors. In this way, it appears that the social conditions of neighborhoods may be important, particularly in places that have experienced declines in the built environment. In addition, we identified evidence that social integration moderates the relationship between neighborhood vacancy and mental health outcomes. The level of neighborhood vacancies had a weaker relationship to wellbeing among those with higher levels of social ties. But none of the independent variables in our study were able to predict social integration, highlighting some potential areas for future research. From these findings, we posit that establishing strong social connections can buffer residents against negative mental health outcomes, and health promotion efforts could usefully assist in maintaining social ties among neighbors.
Keywords: Wellbeing, Vacancy, Mental health, Post-industrial, Social capital, Flint
Introduction
There is now widespread evidence of the relationships between neighborhood conditions and the health of residents. Neighborhood built, social, and physical environments may both bolster and hinder health (Benton et al. 2016; Caspi et al. 2012; Feng et al. 2010; Stevenson et al. 2009; WHO 2016). For example, residents who had longer term residence in their homes had less financial stress and more social support, but higher police-related stress (Schulz et al. 2006). Oh (2003) also echoes the positive association between tenure and social network strength. Generally, less secure housing tenure is associated with poorer health outcomes (Evans et al. 2003), including children’s mental health (Leventhal and Brooks-Gunn 2000). Furthermore, it appears that the type of housing tenure matters for health. Renters have been more likely than homeowners to report issues related to quality of life and a higher mental and physical health burden (Ellaway and Macintyre 1998).
The pathway through which length of residence is thought to influence health is via the development of supportive relationships and trust among neighbors. This form of social integration—including social ties and social support—benefits physical and mental health in urban settings. For example, the level of social integration has been associated with length of time living in a community, with different effects for wealthy and poor neighborhoods. In one study in Chicago, the relationship between length of residence and social integration was stronger in poor neighborhoods compared to wealthy neighborhoods (Keene et al. 2013). The development of strong social ties may be limited in neighborhoods with high levels of mobility or high levels of vacancy. In such declining neighborhoods, vacancy may be more relevant for developing social ties and, thus, may be a more important predictor of mental health than length of residence.
Communities characterized by blighted and vacant properties are also often the same places where social integration may be eroding. Social integration is important because strong ties provide “a sense of belongingness and general social identity, which sociological theorists have argued as being relevant for the promotion of psychological well-being” (Kawachi and Berkman 2001, p463). Conversely, social disintegration is related to increased stress (through a perceived loss of functional support), decreased psychological well-being (through the loss of social networks), an increase in depressive symptoms, and anxiety (Kawachi and Berkman 2001; Oxman et al. 1992).
Neighborhood disorder—manifested as unmaintained vacant properties—also creates disparities in mental health outcomes. Related to length of residence, these outcomes are important because “…disorder’s negative effect on health is greater for long-term residents than for short-term residents” (Hill et al. 2005). These authors report that neighborhood disorder tends to have little impact on recent residents. The perceived characteristics of disorder and disintegration may have also been a precipitating factor of ‘white flight’ (Schulz et al. 2008), a process which has consequently had profoundly negative impacts for predominately African American residents through subsequent public disinvestment (Sadler and Lafreniere 2016). Wallace and Wallace (1990) previously reported on New York City’s decline from the lens of social disintegration. A rash of fires/arsons and home abandonment generated a shift of the population within the city, creating new social challenges for already-disintegrating communities. They suggested that “social disintegration could precipitate psychiatric disorder through interference with…essential striving sentiments” (Wallace and Wallace 1990, p409).
Based on what is understood of the relationships among mental health, length of residence, social disintegration, and neighborhood decline, we hypothesize that social integration within a neighborhood may moderate the relationship between vacancy and mental health. That is, the presence of strong social networks and community social capital may help stem the negative effects of living in a neighborhood that is in decline (having high levels of vacancy).
In Flint, Michigan, a ‘shrinking’ or ‘legacy’ city, the effect of the deterioration of the neighborhood built environment on mental health is mediated by measures of social integration (Kruger et al. 2007). Thus, particularly in post-industrial settings, complex interconnections appear to exist between visible signs of decline in the built features of neighborhoods and the ways in which neighbors interact with one another and perceive their social conditions. Likewise, both of these processes affect mental health, stress, and anxiety. It is possible, however, that strong social ties between neighbors may reduce stress, even in the face of neighborhood decline of the built environment.
Thus, the overall aim of this study was to explore the relationships between length of residence, feelings of social disintegration, vacancy rates, and mental health in a sample of Flint residents. To test these relationships and examine whether the effects of neighborhood vacancy on mental health differ by level of social integration, this study aimed to answer the following questions: 1) What are the independent effects of length of residence, neighborhood vacancy rates, and social disintegration on mental health measures?; 2) Does social disintegration act as a moderator of the relationship between vacancy and mental health?; and 3) What neighborhood characteristics predict social disintegration?
Methods
Study Site
Our study site of Flint, Michigan, has experienced a range of chronic stressors which may influence its residents’ perceptions of mental health and wellbeing. Recently, the water crisis that poisoned hundreds of children (Zahran et al. 2017; Hanna-Attisha et al. 2016) differentially affected poorer, declining neighborhoods (Sadler et al. 2017a)and generated a flurry of concern and activity around resiliency, stress, and trauma (Recast 2017). Prior to that, years of disinvestment through the 1980s to 2000s had already had a negative impact on Flint residents’ ability to live in safe, vibrant neighborhoods with adequate infrastructure to support healthy living. Even before the economic disinvestment by the once-burgeoning auto industry, racially-motivated white flight saw the departure of thousands of families from the city during the 1960s and 1970s (Sadler and Lafreniere 2016). Through the process of deindustrialization, vacant properties, unemployment, drug use, and crime all rose concomitantly (Sadler and Highsmith 2016).
The Genesee County Land Bank has partnered with community organizations to help alleviate some of the adverse consequences of this abandonment (including through beautifying vacant properties). Research has shown this work to have perceived benefits around neighborliness and appearance of the neighborhood (Sadler and Pruett 2015), as well as quantified benefits toward reducing crime (Sadler et al. 2017b). Although this program now maintains between 2000 and 3000 vacant properties every year, it covers only a small fraction of the 24,000 vacant properties in the city (Genesee County Land Bank 2017). Additionally, most of the properties maintained through this program are in highly vacant neighborhoods and not proximate to the participants in our sample. Further and importantly more broadly, the lack of literature linking neighborhood vacancy and integration in the neighborhood with self-reported mental health outcomes leaves an important gap to be filled.
Survey Data
The survey was conducted by the Speak to Your Health Survey Committee, as part of their established bi-annual survey efforts (Kruger et al. 2009). Sampling followed a mixed methods data collection procedure whereby both telephone and in-person surveys were conducted in 2013, with an overall response rate of 39% (Kruger et al. 2017). Telephone surveys followed a random sample of households representing all census tracts in the City of Flint and surrounding area. In-person surveys followed a snowball sampling approach and emphasized oversampling of underrepresented neighborhoods within the City of Flint and surrounding suburbs. Inclusion criteria included being a resident of Genesee County and being 18 years of age or older. All participants provided informed consent prior to participating in the study. We restricted our study’s sample to only those living within the City of Flint with home address data included (n = 260), as vacancy data was only available for neighborhoods within Flint.
1. Mental health measures
The mental health measures used in this study can be categorized as self-reported mental health, hopelessness, and wellbeing. Self-reported mental health was measured using a Likert-type scale (ranging from 1 = excellent mental health to 5 = poor mental health). Hopelessness and wellbeing measures were scales we created using Cronbach’s alpha for survey items, whereby items were first standardized and then those that lowered the alpha score were removed from each scale. Scores were generated by summing standardized items, reversing the scoring for items that had negative correlations with the factor being measured. For both hopelessness and wellbeing, higher scale values indicate higher levels of hopelessness or wellbeing, respectively. Hopelessness items were adapted from the National Survey of Midlife Development in the United States (MIDUS; http://midus.wisc.edu/). For hopelessness, the following agree/disagree statements were included, for a final alpha of 0.795:
I can often tell how things will turn out.
I try to understand how I got into a situation and figure out how to handle it.
I often find the bright side to a bad situation.
I don’t give up until I solve my problems.
I often make plans in advance.
Wellbeing items were adapted from the Brief Symptom Inventory-18 depression subscale (Derogatis 2001) and a global measure of perceived stress (Cohen et al. 1983). For wellbeing, the scale was developed to indicate the opposite of depressive feelings. The following questions about frequency of feelings over the past week were included for a final alpha of 0.875:
How often did you feel nervous or stressed out?
How often did you feel you had so many problems that you didn’t know how to deal with them?
How often did you lonely?
How often did you feel blue or sad?
How often did you feel no interest in things?
2. Neighborhood social disintegration measure
Social disintegration was created as a scale, again using Cronbach’s alpha for survey items on neighborhood attitudes (2008 Boston Neighborhood Survey; (Sampson et al. 1997) and by removing items which lowered the alpha. For this variable, higher scale values indicate being less socially integrated in the neighborhood. The following Likert-type agree/disagree questions were included (1 = strongly agree, 5 = strongly disagree), for a final alpha of 0.844:
How often do people visit one another’s homes?
People are willing to help their neighbors.
People in this neighborhood can be trusted.
I live in a neighborhood where people know each other.
There are adults in my neighborhood that children can look up to.
3. Demographic data and length of time in community
The following variables were obtained during the survey and used in this study: respondent sex, ethnicity (converted to white versus non-white), partner status, education level (seven ordinal categories), age, employment status, and homeownership status. Length of time the respondent resided at the address was measured in months.
Neighborhood Vacancy Data
Based on previous research on neighborhood decline in Flint (Sadler and Lafreniere 2016), we selected the following variables characterizing neighborhood vacancy: long-term change in vacancy rates (1950 vs. 2010) and two short-term changes in vacancy rates (2000 vs. 2010 and 1990 vs. 2000). These time periods were selected for the following reasons. The long-term variable is used to illustrate net neighborhood change from near Flint’s peak population (in the 1950s) to the most recent time period (2010). Our assumption is that neighborhoods that have declined the most over this long-term period are characteristically different from neighborhoods that are either more stable or have more recently started to decline (regardless of when they declined during that period). In some cases, we might expect long-term decline neighborhoods to be more stable because their most severe decline would have occurred at some point in the past (though we do not make distinctions between early and later declines). The short-term variables are included because they indicate a different type of neighborhood: those that were more stable through the tumultuous 1970s and 1980s, but which have more recently seen increases in vacancies. We hypothesize that the driving forces behind decline in these neighborhoods is less a function of economic decline and more a function of neighborhood filtering and racism (these are in contrast to neighborhoods which were blockbusted but supplanted with a replacement population). Thus, we expect the manifestation of this decline on neighborhood social integration will be different than for more historically declining neighborhoods.
Using data from the National Historical Geographic Information System (Manson et al. 2017), vacancy rates were calculated from census tract or block group boundaries for all census years 1950 to 2010 inclusive. Neighborhood was defined as areas with common economic condition factors (ECFs). ECFs are an appraisal term used in past research in Flint to aggregate parcels with similar in age, construction quality, and sale value to neighborhood units (n =102) (Sadler and Lafreniere 2016), and are delineated in Fig. 1. Each respondent was assigned to an ECF based on the address of residence at the time of the survey using ArcGIS v10.3 (ESRI, Redlands, CA, USA). These ECFs were used rather than census units to account for the problem that census tracts and block groups change boundaries over time, thus comparing them from one year to another becomes problematic. Here, we have split census units along ECF boundaries similar to Sadler and Lafreniere (2016) such that: “if x% of a unit fell within a particular ECF, that ECF would receive x% of the values (e.g. population by race, # living in poverty) from that unit” (p. 193).
Fig. 1.
Flint home vacancies per neighborhood (ECFs) by decade, from 1950 to 2010
Statistical Analyses
We first examined descriptive statistics for all variables of interest. For our first research question, we first ran regression models to predict each of our mental health outcomes using demographic predictors alone (age, sex, ethnicity, partnered status, education, employment, and homeownership status). Next, we added the predictors of interest in a stepwise fashion (length of residence, neighborhood vacancy and social disintegration). To address the second research question, we fitted a complete model with all of the above variables plus an interaction term between high vacancy and high social disintegration (using high versus low median values as a cutoff). To examine the third research question, we fitted models to predict neighborhood social disintegration. For models which included area-level predictors (e.g., neighborhood vacancy), we used a multi-level linear regression model to account for clustering at the neighborhood (ECF) level.
Initially, we examined three different temporal scales for neighborhood vacancy, but the long-term (1950s to 2010) and the recent vacancy rates (2000 to 2010) were consistently not significantly associated with any of the outcomes. For this reason, we examined additional time periods that may have had a significant effect on our outcomes of interest (assuming that there may be a lag effect of vacancy such that the process of increasing vacancy itself may be less important than how it fundamentally changes neighborhoods over time). Of each decennial period between 1950 and 2010, only results for the 1990 to 2000 neighborhood vacancy rates were significant; they are shown in the results (other results available upon request). All statistical analyses were conducted using Stata v14 (StataCorp, College Station, TX, USA).
Results
Out of the 260 respondents residing in the City of Flint, the median level of education was more than a high school diploma (Table 1). Most participants were female (66%) and only 37% were white or partnered. On average, participants were 60 years old (median 61 years, sd = 17 years). Twenty-seven percent of participants were unemployed. Most participants (68%) owned their home. We had only minor missing values for education, sex, white ethnicity, partner, and age (up to 5 missing values, not shown). The average scores were 2.4 for self-reported mental health (sd = 1.2), 0.01 for hopelessness (sd = 0.75), 0.001 for wellbeing (sd = 0.92) and < 0.001 for social disintegration (sd = 0.83).
Table 1.
Descriptive statistics for respondent characteristics and characteristics of their neighborhoods
| Individual characteristics | |
|---|---|
| Education‡, mean (sd) | 3.6 (1.9) |
| Age, mean (sd) | 60 (17) |
| Months lived at residence, mean (sd) | 212.6 (197.4) |
| Resided at address less than 5 years, % | 25 |
| Female, % | 66 |
| White, % | 37 |
| Partnered, % | 37 |
| Unemployed, % | 27 |
| Owns home, % | 68 |
| Self-reported mental health score⁑, mean (sd) | 2.4 (1.2) |
| Hopelessness score, mean (sd) | 0.01 (0.75) |
| Wellbeing score, mean (sd) | 0.001 (0.92) |
| Social disintegration score, mean (sd) | < 0.001 (0.83) |
| Characteristics of neighborhoods where participants resided† | |
| Vacancy change 1950–2010, mean (sd) | 0.15 (0.09) |
| Vacancy change 2000–2010, mean (sd) | 0.08 (0.05) |
| Vacancy change 1990–2000, mean (sd) | 0.03 (0.04) |
where 1 = less than high school, 2 = high school graduate or GED, 3 = some college, no degree, 4 = technical school, 5 = associates degree, 6 = bachelor’s degree, 7 = masters, doctorate, or post-doctoral studies
higher values indicate poorer mental health
higher values indicate higher levels of hopelessness
higher values indicate better mental wellbeing
higher values indicate less social integration
higher values indicate higher vacancy rates over the entire time period; negative values imply growth
On average, participants were long-term residents, as they resided in their neighborhood for 212 months, or 17.7 years (median = 13 years, sd = 16 years). Still, 25 % of participants resided in the neighborhood for fewer than five years. On average, participants resided in neighborhoods with 15% higher vacancy from the 1950s to 2010 (median 15%, sd = 9%), 8% higher vacancy from 2000 to 2010 (median 9%, sd = 5%), and 4% higher vacancy from 1990 to 2000 (median 3%, sd = 4%). The spatial distribution of vacancies over time can be described as expanding over time, starting in the central and south eastern neighborhoods in the 1960s and throughout the city by 2010 (with pockets of low vacancies (Fig. 1)).
In answering the question whether there was an effect of length of residence, neighborhood vacancy rates, or social disintegration on mental health measures, we fitted a series of models (Tables 2, 3 and 4). In the models adjusted for demographic covariates only, we detected significant associations between lower levels of mental health and lower education and unemployed status, which remained throughout all models including increasing numbers of independent variables. After including other independent variables of interest, white ethnicity emerged as a consistent, significant, independent predictor of hopelessness (Table 3, β = 0.248 to 0.260, p ≤ 0.05). In evaluating models of increasing complexity, we found that higher neighborhood vacancy was significantly associated with both increased hopelessness and lower wellbeing, even after adjustment for length of residence health (Table 3, β = 2.913, p = 0.039; and Table 4, β = −3.447, p = 0.044). However, after adjustment for social disintegration, these results attenuated to non-significance. In the models including demographic covariates and all independent variables of interest, we detected a significant association between lower wellbeing and higher levels of social disintegration, regardless of length of residence or neighborhood vacancy (Table 4, β = −1.161, p = 0.017).
Table 2.
Regression results for models predicting poor mental health
| β | 95% CI | β | 95% CI | β † | 95% CI | β † | 95% CI | β †§ | 95% CI | |
|---|---|---|---|---|---|---|---|---|---|---|
| Age | 0.002 | −0.008 – 0.011 | 0.003 | −0.006 – 0.013 | 0.006 | −0.003 – 0.016 | 0.006 | −0.003 – 0.016 | 0.006 | −0.004 – 0.015 |
| Female sex | 0.024 | −0.272 – 0.321 | −0.065 | −0.404 – 0.274 | −0.012 | −0.340 – 0.316 | −0.011 | −0.338 – 0.317 | −0.012 | −0.341 – 0.316 |
| White ethnicity | 0.107 | −0.192 – 0.406 | 0.152 | −0.174 – 0.479 | 0.099 | −0.219 – 0.417 | 0.098 | −0.221 – 0.416 | 0.062 | −0.262 – 0.386 |
| Partnered status | 0.138 | −0.151 – 0.426 | 0.154 | −0.171 – 0.479 | 0.177 | −0.136 – 0.490 | 0.186 | −0.130 – 0.501 | 0.200 | −0.117 – 0.517 |
| Education | −0.108 | −0.187 – −0.030 | −0.105 | −0.190 – −0.020 | −0.087 | −0.169 – −0.005 | −0.084 | −0.167 – −0.002 | −0.092 | −0.178 – −0.009 |
| Unemployment | 0.872 | 0.556–1.187 | 0.878 | 0.506–1.250 | 0.930 | 0.574–1.287 | 0.924 | 0.567–1.281 | 0.953 | 0.593–1.312 |
| Home ownership | 0.170 | −0.146 – 0.486 | 0.016 | −0.346 – 0.378 | 0.075 | −0.273 – 0.423 | 0.079 | −0.270 – 0.427 | 0.067 | −0.282 – 0.414 |
| Length of residence | <0.001 | <−0.001–0.001 | <0.001 | <−0.001–0.001 | <0.001 | <−0.001–0.001 | <0.001 | <−0.001–0.001 | ||
| Neighborhood vacancy‡ | 4.073 | −0.167 – 8.313 | 3.987 | −0.270 – 8.244 | −0.215 | −0.695 – 0.264 | ||||
| Social disintegration | 0.044 | −0.136 – 0.225 | 0.044 | −0.374 – 0.461 | ||||||
| High vacancy*High social disintegration | 0.373 | −0.249 – 0.995 |
Bold text = significant at p < 0.05 level
Vacancy = 1990 to 2000
Hierarchical model
High vacancy and high social integration used in this model. Median values set as cut-offs
Table 3.
Regression results for models predicting hopelessness
| β | 95% CI | β | 95% CI | β † | 95% CI | β † | 95% CI | β †§ | 95% CI | |
|---|---|---|---|---|---|---|---|---|---|---|
| Age | 0.003 | −0.003 – 0.008 | 0.004 | −0.002 – 0.001 | 0.004 | −0.002 – 0.001 | 0.004 | −0.002 – 0.010 | 0.004 | −0.002 – 0.010 |
| Female sex | 0.060 | −0.134 – 0.254 | 0.004 | −0.203 – 0.212 | −0.025 | −0.229 – 0.180 | −0.019 | −0.222 – 0.184 | −0.010 | −0.212 – 0.193 |
| White ethnicity | 0.154 | −0.042 – 0.350 | 0.248 | 0.048–0.447 | 0.246 | 0.041–0.451 | 0.240 | 0.036–0.443 | 0.260 | 0.059–0.462 |
| Partnered status | 0.060 | −0.128 – 0.249 | 0.010 | −0.189 – 0.209 | 0.013 | −0.182 – 0.208 | 0.032 | −0.163 – 0.227 | 0.047 | −0.148 – 0.241 |
| Education | −0.096 | −0.147 – −0.045 | −0.079 | −0.131 – −0.027 | −0.067 | −0.119 – −0.015 | −0.062 | −0.114 – −0.011 | −0.065 | −0.116 – −0.013 |
| Unemployment | 0.133 | −0.073 – 0.340 | 0.082 | −0.145 – 0.310 | 0.091 | −0.132 – 0.313 | 0.080 | −0.142 – 0.301 | 0.098 | −0.123 – 0.318 |
| Home ownership | 0.004 | −0.202 – 0.210 | −0.164 | −0.385 – 0.057 | −0.161 | −0.377 – 0.055 | −0.153 | −0.368 – 0.061 | −0.175 | −0.388 – 0.039 |
| Length of residence | <−0.001 | <−0.001 – <0.001 | <−0.001 | <−0.001 – <0.001 | <−0.001 | <−0.001 – <0.001 | <−0.001 | <−0.001 – <0.001 | ||
| Neighborhood vacancy‡ | 2.913 | 0.143–5.683 | 2.673 | −0.089 – 5.435 | 0.041 | −0.257 – 0.339 | ||||
| Social disintegration | 0.102 | −0.011 – 0.214 | −0.045 | −0.301 – 0.211 | ||||||
| High vacancy*High social disintegration | 0.347 | −0.036 – 0.01 |
Bold text = significant at p < 0.05 level
Vacancy = 1990 to 2000
Hierarchical model
High vacancy and high social integration used in this model. Median values set as cut-offs
Table 4.
Regression results for models predicting wellbeing
| β | 95% CI | β | 95% CI | β † | 95% CI | β † | 95% CI | β †§ | 95% CI | |
|---|---|---|---|---|---|---|---|---|---|---|
| Age | −0.003 | −0.010 – 0.003 | −0.003 | −0.010 – 0.004 | −0.002 | −0.009 – 0.005 | −0.002 | −0.009 – 0.005 | −0.002 | −0.009 – 0.005 |
| Female sex | −0.046 | −0.267 – 0.174 | 0.006 | −0.247 – 0.258 | 0.011 | −0.232 – 0.254 | −0.001 | −0.241 – 0.239 | −0.010 | −0.247 – 0.227 |
| White ethnicity | −0.012 | −0.235 – 0.210 | −0.035 | −0.277 – 0.208 | −0.063 | −0.308 – 0.182 | −0.047 | −0.288 – 0.193 | −0.077 | −0.317 – 0.166 |
| Partnered status | −0.030 | −0.245 – 0.185 | −0.116 | −0.359 – 0.128 | −0.104 | −0.337 – 0.129 | −0.137 | −0.369 – 0.094 | −0.166 | −0.397 – 0.064 |
| Education | 0.043 | −0.015 – 0.101 | 0.054 | −0.010 – 0.117 | 0.037 | −0.025 – 0.099 | 0.030 | −0.031 – 0.092 | 0.030 | −0.031 – 0.091 |
| Unemployment | −0.350 | −0.585 – −0.115 | −0.421 | −0.698 – −0.144 | −0.427 | −0.693 – −0.162 | −0.414 | −0.676 – −0.153 | −0.441 | −0.701 – −0.181 |
| Home ownership | 0.088 | −0.145 – 0.322 | 0.132 | −0.136 – 0.399 | 0.136 | −0.118 – 0.390 | 0.122 | −0.129 – 0.373 | 0.151 | −0.097 – 0.399 |
| Length of residence | <−0.001 | <−0.001 – <0.001 | <−0.001 | <−0.001 – <0.001 | <−0.001 | <−0.001 – <0.001 | <0.001 | <−0.001 – <0.001 | ||
| Neighborhood vacancy‡ | −3.447 | −6.794 – −0.099 | −3.069 | −6.349 – 0.211 | 0.001 | −0.360 – 0.363 | ||||
| Social disintegration | −0.161 | −0.294 – −0.029 | −0.086 | −0.388 – 0.216 | ||||||
| High vacancy*High social disintegration | −0.416 | −0.867 – 0.035 |
Bold text = significant at p < 0.05 level
Vacancy = 1990 to 2000
Hierarchical model
High vacancy and high social integration used in this model. Median values set as cut-offs
To answer the second research question, we first fitted regression models of the relationship between vacancy and health, stratified by high versus low social disintegration status, examined a fully adjusted model, and included an interaction term for high vacancy*high social disintegration (results not shown). We found a significant, negative relationship between vacancy and wellbeing among those with high social disintegration and a null association among those with low social disintegration. In our fully adjusted models including the interaction term (final model in Tables 2, 3 and 4), we found neither vacancy, social disintegration, nor length of residence were significantly associated with any of the three mental health measures. It is worth noting, however, that the interaction term was approaching significance for both hopelessness (p = 0.076) and wellbeing (p = 0.071), with coefficients in the expected directions. Unemployment remained significantly associated with both poorer mental health (Table 2, β = 0.953, p < 0.001) and lower wellbeing (Table 4, β = −0.441, p = 0.002). Lower education remained significantly associated with poorer mental health (Table 2, β = −0.092, p = 0.029) and higher levels of hopelessness (Table 3, β = −0.065, p = 0.018).
Last, we did not detect a significant association between length of residence and social disintegration (Table 5). In fact, none of the independent variables in this study were significant predictors of social disintegration.
Table 5.
Regression results for models predicting social disintegration
| β | 95% CI | β† | 95% CI | β† | 95% CI | |
|---|---|---|---|---|---|---|
| Length of residence | <−0.001 | <−0.001 – <0.001 | <−0.001 | <−0.001 – <0.001 | ||
| Neighborhood vacancy± | 2.33 | −0.64 – 5.29 | 2.45 | −1.12 – 6.02 | ||
| Age | 0.003 | −0.01 – 0.10 | 0.002 | −0.01 – 0.01 | 0.003 | −0.01 – 0.01 |
| Female sex | −0.02 | −0.28 – 0.23 | 0.02 | −0.20 – 0.24 | −0.06 | −0.32 – 0.19 |
| White ethnicity | −0.03 | −0.28 – 0.22 | 0.05 | −0.18 – 0.28 | 0.08 | −0.18 – 0.34 |
| Partnered status | −0.20 | −0.44 – 0.05 | −0.13 | −0.35 – 0.08 | −0.18 | −0.42 – 0.06 |
| Education | −0.05 | −0.11 – 0.02 | −0.04 | −0.10 – 0.02 | −0.04 | −0.11 – 0.02 |
| Unemployment | 0.16 | −0.12 – 0.45 | 0.13 | −0.10 – 0.37 | 0.10 | −0.18 – 0.38 |
| Home ownership | −0.06 | −0.34 – 0.21 | −0.001 | −0.24 – 0.23 | −0.08 | −0.35 – 0.18 |
Vacancy= 1990 to 2000
Hierarchical model
Discussion
In this study, we sought to uncover associations between mental health/wellbeing and neighborhood conditions in a post-industrial city. We found that short-term (not long-term) increases in neighborhood vacancy in the recent past were associated with poorer mental health. What is noteworthy is that not only were the most recent changes in vacancy significant predictors of mental health, indeed, our most significant results came from a time period that concluded 13 years prior to the survey data. We hypothesize that the effects of neighborhood change are not immediate and direct. Rather, a neighborhood with declining population in one time-period may experience a time lag effect such that, after the neighborhood loses population, a further decline in neighborhood conditions that bolster mental health occurs (e.g., neighborhood friends, support, good housing, amenities and public services) which, in turn, affects mental health. When neighborhood and individual covariates and an interaction term were included, however, we did not detect significant associations between our independent variables of interest and the three mental health measures. The interaction term between social disintegration and vacancy rate was approaching significance for both hopelessness and wellbeing, even with our relatively small sample size. In the hopes that these preliminary findings may be tested in other post-industrial cities, using larger samples, we feel this finding warrants discussion. However, future research would benefit from examining similar relationships, in a larger sample.
From this study, it appears that there is a negative health effect of a combination of a lack of social ties in neighborhoods that have higher levels of vacancy. Among people with higher levels of social ties to neighbors, the effect of vacancy on mental health is null. Thus, social integration may offer a protective buffer to mental health in post-industrial cities with high vacancy rates. In this way, it appears that the social conditions of a neighborhood may be important predictors of mental wellbeing, even in (or particularly in) places that have experienced declines in the built environment. In another study in Flint, the number of dilapidated residential buildings near the home was associated with depressive symptoms and stress, with apparent mediation by social contact with neighbors (Kruger et al. 2007). Similar to our findings, social contact was protective against poorer mental health (Kruger et al. 2007), with evidence of social capital even in the most deteriorating neighborhoods (Sadler and Pruett 2015). We also did not see evidence that neighborhood social integration depends on length of residence. Interestingly, none of the independent variables in our study were able to predict social integration, highlighting some other potential areas for future research. We did find that education and employment are important predictors of mental health, regardless of neighborhood conditions and other factors, which echoes other work (Dooley 2003; Hampson et al. 2007). Efforts to improve mental health should also consider these individual-level factors in health promotion.
Within Flint, property abandonment has not abated. Our finding that recently vacated neighborhoods have poorer mental health outcomes is, however, hopeful from a long-term perspective. Should Flint continue to decline in population, the negative effects that neighborhoods experience by transitioning away from dense urban neighborhoods may abate as they move toward lower-density ‘green’ neighborhoods. Some anecdotal evidence of this exists in Flint, where neighborhoods that declined rapidly 30 years ago have now stabilized and are experiencing residential and commercial reinvestment.
Flint benefits from a recent master planning effort that deeply engaged the community throughout the process and sought to eliminate notions of displacement in favor of working to maintain varying typologies of neighborhoods (e.g. lower-density ‘green’ neighborhoods versus traditional neighborhoods of typical density). Our findings suggest that efforts to promote social integration in neighborhoods among the remaining residents (regardless of the level of vacancy) may promote mental wellbeing.
This study has limitations. First, the survey recruitment involved a mix of random sampling and snowball sampling for hard-to-reach neighborhoods. As such, care should be taken in terms of generalizability of these findings. The majority of respondents were older and sample size was limited. We only include one objective measure of the built environment—residential vacancy—which was assessed by NHGIS rather than on-the-ground assessment. In addition, several other important aspects of the built environment have been shown to affect mental health in post-industrial settings, including evidence of arson, broken windows, graffiti, and litter (Robinson et al. 2003). All our measures of mental health and social conditions were self-reported. As mentioned, future studies on large samples are recommended to either confirm or disconfirm our findings. Future research could also usefully further explore connections between built environmental factors (e.g., crime, objectively measured blight) and neighborhood social integration. Alternatively, because we did not identify any independent predictors of social integration, future qualitative research could provide useful insight into meanings of social integration in post-industrial cities and how this might differ from other places.
Our results are notable because they demonstrate that even in Flint—where neighborhood decline is extreme and highly variable—vacancy alone is not responsible for variations in mental health. Rather, social support and ties between neighbors can buffer residents from the ill mental health effects of vacancy.
Conclusions
We posit that establishing strong social connections can buffer residents against negative mental health outcomes and deliberate efforts to help maintain social ties may yield the best results. Such findings may be useful for community/grassroots organization efforts throughout the Rust Belt region, particularly given the challenges of managing abundant vacant land in cities such as Flint. Even so, these findings should not detract from urban planning and reinvestment strategies and efforts to provide safe opportunities for health equitably across the city (e.g., blight removal, healthy food options). As Rust Belt cities see their sharpest population declines even out, the challenge moving forward will be rebuilding these cities in ways that are more equitable than the landscapes left behind; understanding the how social connectedness can buffer against the negative influences of the built environment will help create such equitable places. Our work does not ignore the challenges of promoting social connectedness in communities marked by decline. Rather, by advocating for the mechanisms by which residents in such neighborhoods can build social connectedness and the potential role of social connectedness in mental health, we hope such mechanisms can be promoted in other communities. Ultimately, as these cities continue to evolve (from industrial, to post-industrial, to the next phase), we will need continued evidence to assess what elements may be necessary to promote health and equity.
Acknowledgements
The authors wish to acknowledge Amanda Rzotkiewicz for her assistance with manuscript components.
Footnotes
Compliance with Ethical Standards
Conflict of Interest The authors have no conflicts to declare.
References
- Bank GCL (2017). Vacant property maintenance plan. Genesee County Land Bank. [Google Scholar]
- Benton JS, Anderson J, Hunter RF, & French DP (2016). The effect of changing the built environment on physical activity: A quantitative review of the risk of bias in natural experiments. International Journal of Behavioral Nutrition and Physical Activity, 13, 107. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Caspi CE, Sorensen G, Subramanian SV, & Kawachi I (2012). The local food environment and diet: A systematic review. Health & Place, 18, 1172–1187. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Cohen S, Kamarck T, & Mermelstein R (1983). A global measure of perceived stress. Journal of Health and Social Behavior, 24, 385–396. [PubMed] [Google Scholar]
- Derogatis LR (2001). Brief symptom inventory-18 (bsi-18) administration, scoring, and procedures manual. NCS Pearson: Minneapolis. [Google Scholar]
- Dooley D (2003). Unemployment, underemployment, and mental health: Conceptualizing employment status as a continuum. American Journal of Community Psychology, 32, 9–20. [DOI] [PubMed] [Google Scholar]
- Ellaway A, & Macintyre S (1998). Does housing tenure predict health in the UK because it exposes people to different levels of housing related hazards in the home or its surroundings? Health & Place, 4, 141–150. [DOI] [PubMed] [Google Scholar]
- Evans GW, Wells NM, & Moch A (2003). Housing and mental health: A review of the evidence and a methodological and conceptual critique. Journal of Social Issues, 59, 475–500. [Google Scholar]
- Feng J, Glass TA, Curriero FC, Stewart WF, & Schwartz BS (2010). The built environment and obesity: A systematic review of the epidemiologic evidence. Health & Place, 16, 175–190. [DOI] [PubMed] [Google Scholar]
- Hampson SE, Goldberg LR, Vogt TM, & Dubanoski JP (2007). Mechanisms by which childhood personality traits influence adult health status: Educational attainment and healthy behaviors. Health Psychology, 26, 121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hanna-Attisha M, Lachance J, Sadler RC, & Champney Schnepp A (2016). Elevated blood lead levels in children associated with the Flint drinking water crisis: A spatial analysis of risk and public health response. American Journal of Public Health, 106, 283–290. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hill TD, Ross CE, & Angel RJ (2005). Neighborhood disorder, psychophysiological distress, and health. Journal of Health and Social Behavior, 46, 170–186. [DOI] [PubMed] [Google Scholar]
- Kawachi I, & Berkman LF (2001). Social ties and mental health. Journal of Urban Health, 78, 458–467. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Keene D, Bader M, & Ailshire J (2013). Length of residence and social integration: The contingent effects of neighborhood poverty. Health & Place, 21, 171–178. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kruger DJ, Reischl TM, & Gee GC (2007). Neighborhood social conditions mediate the association between physical deterioration and mental health. American Journal of Community Psychology, 40, 261–271. [DOI] [PubMed] [Google Scholar]
- Kruger DJ, Shirey LA, Morrel-Samuels S, Skorcz S, & Brady JS (2009). Using a community-based health survey as a tool for informing local health policy. Journal of Public Health Management and Policy, 15, 47–53. [DOI] [PubMed] [Google Scholar]
- Kruger DJ, Cupal S, Franzen SP, Kodjebacheva G, Bailey ES, Key KD, & Kaufman MM (2017). Toxic trauma: Household water quality experiences predict posttraumatic stress disorder symptoms during the Flint, Michigan, water crisis. Journal of Community Psychology, 45, 957–962. [Google Scholar]
- Leventhal T, & Brooks-Gunn J (2000). The neighborhoods they live in: The effects of neighborhood residence on child and adolescent outcomes. Psychological Bulletin, 126, 309. [DOI] [PubMed] [Google Scholar]
- Manson S, Schroeder J, van Riper D, & Ruggles S (2017). IPUMS national historical geographic information system: Version 12.0. Minneapolis: University of Minnesota. [Google Scholar]
- Oh J-H (2003). Assessing the social bonds of elderly neighbors: The roles of length of residence, crime victimization, and perceived disorder. Sociological Inquiry, 73, 490–510. [Google Scholar]
- Oxman TE, Berkman LF, Kasl S, Freeman JDH, & Barrett J (1992). Social support and depressive symptoms in the elderly. American Journal of Epidemiology, 135, 356–368. [DOI] [PubMed] [Google Scholar]
- Recast F (2017). Flint Resiliency in Communities After Stress and Trauma Program [Online]. Available: https://www.facebook.com/FlintReCAST/ [Accessed].
- Robinson JB, Lawton BA, Taylor RB, & Perkins DD (2003). Multilevel longitudinal impacts of incivilities: Fear of crime, expected safety, and block satisfaction. Journal of Quantitative Criminology, 19, 237–274. [Google Scholar]
- Sadler RC, & Highsmith AR (2016). Rethinking Tiebout: The contribution of political fragmentation and racial/economic segregation to the Flint water crisis. Environmental Justice, 9, 143–151. [Google Scholar]
- Sadler RC, & Lafreniere DJ (2016). Racist housing practices as a precursor to uneven neighborhood change in a post-industrial city. Housing Studies, 32(2), 186–208. [Google Scholar]
- Sadler RC & Pruett NK (2015). Mitigating blight and building community pride in a legacy city: lessons learned from a land bank’s clean and green programme. Community Development Journal, bsv052. [Google Scholar]
- Sadler RC, Lachance J, & Hanna-Attisha M (2017a). Social and built environmental correlates of predicted blood lead levels in the Flint water crisis. American Journal of Public Health, 107, 763–769. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sadler RC, Pizarro J, Turchan B, Gasteyer SP, & Mcgarrell EF (2017b). Exploring the spatialtemporal relationships between a community greening program and neighborhood rates of crime. Applied Geography, 83, 13–26. [Google Scholar]
- Sampson RJ, Raudenbush SW, & Earls F (1997). Neighborhoods and violent crime: A multilevel study of collective efficacy. Science, 227, 918–923. [DOI] [PubMed] [Google Scholar]
- Schulz AJ, Israel BA, Zenk SN, Parker EA, Lichtenstein R, Shellman-Weir S, & Ab LK (2006). Psychosocial stress and social support as mediators of relationships between income, length of residence and depressive symptoms among African American women on Detroit’s eastside. Social Science & Medicine, 62, 510–522. [DOI] [PubMed] [Google Scholar]
- Schulz AJ, Zenk SN, Israel BA, Mentz G, Stokes C, & Galea S (2008). Do neighborhood economic characteristics, racial composition, and residential stability predict perceptions of stress associated with the physical and social environment? Findings from a multilevel analysis in Detroit. Journal of Urban Health, 85, 642–661. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Stevenson A, Pearce J, Blakely T, Ivory V, & Witten K (2009). Neighbourhoods and health: A review of the New Zealand literature. New Zealand Geographer, 65, 211–221. [Google Scholar]
- Wallace R, & Wallace D (1990). Origins of public health collapse in new York City: The dynamics of planned shrinkage, contagious urban decay and social disintegration. Bulletin of the New York Academy of Medicine, 66, 391–434. [PMC free article] [PubMed] [Google Scholar]
- WHO. (2016). Urban green spaces and health: A review of the evidence. Copenhagen: World Health Organization. [Google Scholar]
- Zahran S, Mcelmurry SP, & Sadler RC (2017). Four phases of the Flint water crisis: Evidence from blood lead levels in children. Environmental Research, 157, 160–172. [DOI] [PMC free article] [PubMed] [Google Scholar]