Identifying county characteristics associated with resident well-being: A population based study
Identifying county characteristics associated with resident well-being: A population based study
Brita Roy 0 1
Carley Riley 1
Jeph Herrin 1
Erica S. Spatz 1
Anita Arora 1
Kenneth P. Kell 1
John Welsh 1
Elizabeth Y. Rula 1
Harlan M. Krumholz 1
0 Department of Internal Medicine, Section of General Internal Medicine, Yale School of Medicine, New Haven, Connecticut, United States of America, 2 Department of Pediatrics, University of Cincinnati College of Medicine, Cincinnati, Ohio, United States of America, 3 Division of Critical Care, Cincinnati Children's Hospital Medical Center , Cincinnati , Ohio, United States of America, 4 Department of Internal Medicine, Section of Cardiovascular Medicine, Yale School of Medicine, Center for Outcomes Research and Evaluation, New Haven, Connecticut, United States of America, 5 Department of Internal Medicine, Section of General Internal Medicine, Yale School of Medicine, New Haven, Connecticut, United States of America , 6 Tivity Health, Franklin , Tennessee, United States of America, 7 Yale University , New Haven , Connecticut, United States of America, 8 Department of Internal Medicine, Section of Cardiovascular Medicine, Yale School of Medicine; Department of Health Policy and Management, Yale School of Public Health; Center for Outcomes Research and Evaluation, Yale-New Haven Hospital , New Haven, Connecticut , United States of America
1 Editor: Michael L. Goodman, University of Texas Medical Branch at Galveston , UNITED STATES
Data Availability Statement: Most data on
community factors used in this study is publicly
available through the Robert Wood Johnson
County Health Rankings (http://www.
rankings-data), Area Health Resource File (https://
0), and the American Community Survey/US
data-tables-and-tools/). Data was also purchased
from Claritas (formerly Nielsen) Pop-Facts, and can
Well-being is a positively-framed, holistic assessment of health and quality of life that is
associated with longevity and better health outcomes. We aimed to identify county attributes
that are independently associated with a comprehensive, multi-dimensional assessment of
We performed a cross-sectional study examining associations between 77 pre-specified
county attributes and a multi-dimensional assessment of individual US residents'
wellbeing, captured by the Gallup-Sharecare Well-Being Index. Our cohort included 338,846
survey participants, randomly sampled from 3,118 US counties or county equivalents.
We identified twelve county-level factors that were independently associated with individual
well-being scores. Together, these twelve factors explained 91% of the variance in
individual well-being scores, and they represent four conceptually distinct categories: demographic
(% black); social and economic (child poverty, education level [<high school, high school
diploma/equivalent, college degree], household income, % divorced); clinical care (%
eligible women obtaining mammography, preventable hospital stays per 100,000, number of
be obtained through their website (http://www.
tetrad.com/demographics/usa/claritas/) or by
calling 1-800-866-6511. In addition, we have
posted a de-identified data set including all
variables included in our study on ICSPR Open.
This data set includes county resident well-being
data from Gallup-Sharecare. The ICSPR site is
publicly available and will allow any researcher to
replicate our results. The data set can be found at
the following DOI: http://doi.org/10.3886/
Funding: This project was primarily supported by
grant number K12HS023000 from the Agency for
Healthcare Research and Quality. The content is
solely the responsibility of the authors and does
not necessarily represent the official views of the
Agency for Healthcare Research and Quality.
Authors of this work were also partially supported
by the Robert Wood Johnson Foundation Clinical
Scholars Program, Veterans Administration, Yale
Center for Clinical Investigation through the Clinical
and Translational Science Award, Healthways, Inc.
(acquired by Sharecare, Inc.), Centers for Medicare
& Medicaid Services, Medtronic, and the Food and
Drug Administration during the time of the study
period. Advisory fees were also received from
UnitedHealth, IBM Watson Health Life Sciences
Board, Element Science, and Aetna during the time
of the study period. Though uncompensated, HMK
is the founder of Hugo. BR and CR were partially
supported by the Institute for Healthcare
Improvement after the completion of the study
period. The funders had no role in study design,
data collection and analyses, decision to publish, or
preparation of the manuscript.
Competing interests: The authors declare the
following competing interests: partial support from
the Agency for Healthcare Research and Quality
(BR and ES) and the Robert Wood Johnson
Foundation (BR, CR, and AA) and the Veterans
Administration (BR) for the submitted work; AA
was partially supported by the Yale Center for
Clinical Investigation through Clinical and
Translational Science Award during the study
period; BR and CR are consultants for the Institute
for Healthcare Improvement 100 Million Healthier
Lives initiative; ER and KK are current or former
employees and shareholders of the Healthways
corporation (acquired by Sharecare), the company
that developed the measure of well-being used in
this article; ES, JH, and HK also report receiving
support from the Centers for Medicare & Medicaid
Services; Dr. Krumholz is a recipient of research
agreements from Medtronic and Johnson &
Johnson (Janssen), through Yale, to develop
methods of clinical trial data sharing; is the
federally qualified health centers); and physical environment (% commuting by bicycle and
by public transit).
Twelve factors across social and economic, clinical care, and physical environmental
county-level factors explained the majority of variation in resident well-being.
Well-being is defined as ªa person's cognitive and affective evaluations of his or her life,º and
includes ªemotional reactions to events as well as cognitive judgements of satisfaction and
] It is a holistic, positively framed assessment of health and quality of life that
captures aspects of an individual's physical, mental, and social health, beyond the presence or
absence of disease, as well as an introspective evaluation of one's life.[2±6] Higher well-being
by definition has inherent positive value, and an individual's well-being has been
independently associated in a dose-response manner with lower risk of cardiovascular events and
greater longevity.[7±12] Recently, our group extended these findings by reporting a
relationship between aggregate well-being and life expectancy at the county level, after accounting for
poverty, education and race.[
] Given the positive value of well-being and the recent focus on
population health, there is interest in designing communities that support this construct,[14±
16] but more evidence is needed about local factors that are associated with population
Where one lives may be an important determinant of a person's well-being, and indeed,
well-being varies by region.[
] Community attributes such as access to basic healthcare and
social services, safe and clean streets, public transportation, green spaces and healthy foods,
along with social aspects such as neighborhood cohesion and trust, may influence peoples'
evaluation of their own well-being.[
] Many of these community attributes are modifiable,
and thus are increasingly recognized as a focus for improving a population's health and
wellbeing. In fact, the Centers for Medicare and Medicaid Services have begun to pilot accountable
health community models.[
] These new payment models acknowledge that social and
community factors influence health outcomes, and compensation to support positive modification
of these factors is needed. Although several community factors have been associated with
wellbeing or aspects of well-being, knowing which of these factors are most strongly associated
with health and well-being is important yet challenging because many of these attributes are
correlated with each other. As such, it is necessary to identify community attributes that are
strongly and independently associated with better health outcomes and higher well-being.
Identifying these attributes could provide evidence for prioritization of community-level tar
gets for interventions that aim to promote well-being.
We used a systematic method[
] to identify independent associations with well-being
among highly interrelated county characteristics drawn from multiple community sectors
emphasized by established theoretical frameworks.[15, 21±25] We performed a cross-sectional
study using data from the Gallup-Sharecare Well-Being Index (WBI), a comprehensive,
multidimensional well-being assessment of over 350,000 Americans annually. We demonstrated
independent associations among individual well-being scores and 77 county factors related to
the following community sectors: demographics, social and economic status, clinical care,
health behaviors, and the physical environment. To compare our results with the emerging
literature related to life satisfaction, we also conducted an exploratory analysis of the relationship
2 / 18
between county factors and life evaluation, a component of the WBI consisting of a subjective
overall assessment of one's current and five-year outlook on life.[
Materials and methods
We used WBI data from January 1, 2010 to December 31, 2012 to assess well-being. Gallup
Sharecare conducted between 500 to 1000 telephone surveys with a random sample each day,
350 days per year. The sample included adults aged 18 years and older residing in the United
States, who spoke either English or Spanish and have either a landline or cellular phone. Gal
lup-Sharecare used a structured sampling design to obtain data from all 50 states and the
District of Columbia.[
The WBI was developed based on the work of psychology experts.[
] Briefly, survey
items that aligned with previous research on well-being were initially compiled by experts in
] Based on reviews of the literature, items were selected to encompass both
hedonic well-being (i.e. , people's feelings and thoughts about their lives) [
] and eudemonic
well-being (i.e., an individual's judgments about the meaning and purpose in one's life),[
and thus incorporated items assessing daily emotional experience and a wide variety of
evaluative domains, such as overall life, standard of living, and satisfaction with community, work,
relationships, and personal health. Factor analysis using data from a large, representative
national sample was then used to determine the final set of questions. Criterion validity of
regionally aggregated data was established by examining correlations with regional health and
] Subsequently, principal component analyses and confirmatory
factor analyses were used to create an instrument valid for measuring well-being at the
individual level. The individual well-being measure has acceptable reliability, internal and external
] It includes 40 self-reported items organized into six domains representing key
aspects of well-being that are similar to other multi-dimensional constructs of well-being[
(S1 Table): life evaluation, emotional health, work environment, physical health, healthy
behaviors, and basic access.[
] Our primary outcome was the composite individual
wellbeing score (iWBS), which is the unweighted mean of the six domain scores, each scaled to
range from 0±100.
Our secondary outcome was the life evaluation index (LEI). The LEI is a two-item measure
adapted from Cantril's Self-Anchoring Striving Scale,[
] a reflection on one's overall
experience of life and future outlook. These items require respondents to rate on a scale from 0 to 10
their overall current life situation as well as their expected life situation in five years. The LEI is
calculated by averaging the number of points (e.g., where the participant places themselves on
the ladder) from both items and multiplying by ten, to match the scale of the other domains
with a maximum score of 100.
Measures: County attributes
We used U.S. county or county equivalent (e.g., boroughs, towns, or parishes, depending on the state) as our unit of analysis because it is a defined unit for community initiatives and policy change. County was the smallest geographic unit for which we could both assign respondents and identify a large range of local attributes.
Existing theoretical frameworks agree that a variety of community properties affect mortal
ity, health-related quality of life, and subjective health and well-being.[
22, 23, 37, 38
After reviewing multiple theoretical frameworks, we adapted and expanded the University of
Wisconsin Population Health Institute County Health Rankings and Roadmaps (CHRR) framework to encompass the major themes of categories of county factors theorized to be
3 / 18
associated with health and well-being across all frameworks. These included social and
economic, health behaviors, clinical care, physical environment, demographic, and psychosocial
categories of county factors. We identified, a priori, 114 county factors within these categories
that we postulated would influence the six domains of resident well-being in our conceptual
model (S1 Fig).
We then searched for zip code or county level sources for data on these 114 pre-specified
county factors hypothesized to be associated with resident well-being. We first drew factors
from the Robert Wood Johnson Foundation County Health Rankings (years 2010, 2012 2013),
which aggregates data from multiple sources. We captured data on additional factors from the
Area Health Resource File (2007±2012), American Community Survey/US Census (2010 and
2012), and Nielsen Pop-Facts (2013). We were able to obtain data on 77 county factors within
5 categories that were well-aligned with the pre-specified factors (Table 1). No consistent
county or zip code level data were available across years 2010±2013 for the other 37 factors (S1
Fig), which included all factors in the psychosocial category, so this category was removed
from our model. Continuous variables were categorized into quintiles by county, unless highly
skewed (i.e., more than 20% of counties having the same value), in which case two observers
inspected the distribution and agreed on reasonable categories. All data were merged with the
WBI participant data using the Federal Information Processing Standard code.
We initially performed descriptive analyses of community attributes across counties. Because
we expected that many county factors would be correlated within and across categories (e.g.,
high percent of population uninsured may be associated with low income; low educational
attainment may be associated with high crime rates), we used a sequential approach to identify
independent factors within each category and then across all categories. First, we evaluated
bivariate associations between each of the 77 county-level predictor variables with participants'
WBI scores using a mixed effects linear model with random effect for county and individual
well-being as the dependent variable. For each model, we calculated the overall (Wald) P-value
for the county attribute and the model R2 as the proportion of variance explained at the county
] We eliminated variables that were not significantly associated with the iWBS (overall
P-value >0.05) or did not explain at least 20% of the variance in the iWBS (R2<0.20). This last
criterion was chosen based on the distribution of R2 values, which had a substantial gap below
20%. Of those variables retained, we assessed for multi-collinearity within each category using
variance decomposition proportions and eliminated redundant measures: if the retained
variables in a category had a singular value greater than 20, variance decomposition proportions
for each variable were examined, and if two or more contributed more than 50%, we retained
the one with the greatest value. This approach has been used previously by health services
researchers to reduce large numbers of related factors to a smaller representative set.[
We then estimated a single model for each category including factors retained from this pro
cess. Then, we estimated a combined model including all variables that were independently
significant (p<0.05) in their respective category-specific model. Our final model included only
variables that were independently significant in the combined model (p<0.05).
Secondary analyses were performed using the same sequential approach, but with the LEI
as the outcome. We focused on this particular domain to separate findings from those related
directly to health behaviors and outcomes, and to compare our results with prior studies that
have used this measure or a similar assessment of overall life evaluation.[
Analyses were performed using Stata 14.1 (2015, StataCorp, College Station, TX). The Yale
University Human Subjects Committee approved this study.
4 / 18
71.4 73.1 73.8
73.1 73.8 74.0
73.9 74.0 73.8
73.7 74.0 Ð
74.1 71.9 Ð
71.7 71.6 72.3
Social and Economic Factors
Each factor was categorized by equally distributed quintiles, unless noted in parentheses. Bivariate associations for each county factor with resident well-being were
tested and level of significance is noted by the Wald P-value. Abbreviations: HbA1c = Hemoglobin A1c; ED = Emergency department; GP = General practitioner;
NH = Nursing home; RN = Registered nurse; BSN = Bachelor of science in nursing; DDS = Doctor of dental surgery; DO = Doctor of osteopathy; MD = Doctor of
The cohort included 338,846 survey participants, representing 3,118 counties in our analysis.
Overall, the mean (SD) iWBS was 73.8 (15.4) and mean LEI was 72.5 (18.8). Counties had a
mean aggregate composite well-being score of 66.5 (SD 5.4), with a range of 39.9 to 90.6.
In bivariate analyses, 73 variables across all categories were significantly associated with
iWBS (p<0.05; Table 1). In the demographic category, measures of race, including higher
percent Black (p = 0.005; R2 = 0.54) and higher percent Asian (p<0.001; R2 = 0.64), as well as
measures of area characteristics, including being a retirement destination (p = 0.003; R2 =
0.54) and being less rural (p<0.001; R2 = 0.59), were associated with higher iWBS, and each
explained over half of the variance in iWBS. In the social and economic category, lower
percent without a high school diploma and higher percent with a Bachelor's degree were both
associated with higher well-being (p<0.001), and these factors explained the greatest amount
of variation in resident well-being (R2 = 0.69 for both). In the clinical care category, variables
related to healthcare spending, including lower percent with Medicaid (p<0.001; R2 = 0.41)
and lower percent income spent on prescription medications (p<0.001; R2 = 0.42) and on
health care (p<0.001; R2 = 0.42), as well as variables related to healthcare capacity, including
greater number of hospitals (p<0.001; R2 = 0.56), number of psychiatric hospitals (p<0.001;
R2 = 0.55), number of federally qualified health centers (p<0.001; R2 = 0.54), and presence of
medical (p<0.001; R2 = 0.55), dental (p = 0.002; R2 = 0.54), pharmacy (p<0.001; R2 = 0.55),
and nursing schools (p<0.001; R2 = 0.54) were associated with higher iWBS scores and
explained the greatest amount of variation in iWBS. In the physical environment category,
higher percent commuting by bicycle (p<0.001; R2 = 0.61) or public transport (p<0.001; R2 =
0.59) or working from home (p<0.001; R2 = 0.60) were associated with higher iWBS scores
and explained the greatest amount of variation in iWBS. In the health behavior category,
number of recreational facilities per hundred thousand explained the greatest amount of variance
in iWBS (p<0.001; R2 = 0.29).
Of the 73 variables significantly associated with iWBS, 40 explained greater than 20% of the
variation in well-being (R2>0.20), and thus were retained for within-category multivariable
analyses (Table 2). In these multivariable models, 14 variables were no longer independently
associated with iWBS after accounting for the effect of other factors in the same category. The
social and economic category explained the greatest amount of variance (R2 = 0.86) and the
health behavior category explained the least amount of variance (R2 = 0.29) in iWBS.
After dropping non-significant variables within each category model, 27 county factors
remained and were included in a combined model assessing county factors with iWBS
(Table 3, Combined Model). Of these, only 13 county factors remained significantly associated
with iWBS after accounting for the effect of other variables in the model and were moved
forward into the final model. In this final model, twelve variables, originally from four different
categories, remained significantly associated with well-being (Table 3, Final Model). None of
7 / 18
the variables from the health behavior category contributed independently to the final model.
Together, these twelve variables explained 91% of the variance in iWBS. The three variables
with the biggest effect on higher individual well-being, as measured by the magnitude of
explained variance lost if the variable were omitted from the full model (Fig 1), were: lower
rates of preventable hospital stays, lower percent divorced, and lower percent without high
In secondary bivariate analyses with LEI as the outcome, almost all county characteristics in
all five categories were significantly associated with the life evaluation scores and 39 of these
variables explained greater than 20% of the variation in the LEI, (S2 Table). None of the
variables in the health behavior category explained greater than 20% variation in LEI, thus no
category-specific model was performed for the health behavior category. The social and economic
category explained the greatest amount of variance (R2 = 0.67), the demographic and physical
environment categories explained a similar amount of variance (R2 = 0.55 and 0.54,
respectively), and the clinical care category explained the least amount of variance in life evaluation
(R2 = 0.49) (S3 Table). The final model included 10 county attributes from the demographic,
social and economic, clinical care, and physical environment categories, and together, these
ten attributes explained 80% of the variance in the LEI (Table 4). The variables with the largest
contribution to R2 were: higher percent black race, lower percent with less than a high school
diploma, and lower percent commuting by public transportation.
In this nationwide study of more than 300,000 adults and more than 75 attributes of the
counties in which they reside, we identified twelve county factors that were independently
associated with a comprehensive, multi-dimensional assessment of individual well-being. Together,
these twelve factors explained more than ninety percent of the variance in individual
wellbeing scores. The final set of twelve factors were from the demographic, social and economic,
clinical care, and physical environment categories. The majority of these factors were also
independently associated with overall life evaluation. These findings suggest that promotion of
diversity as well as targeted investments in education, transportation, and primary care may
lead to higher well-being of community residents, an idea worth testing. Our findings also
bolster existing theoretical models that propose multi-pronged efforts to improve community
9 / 18
ref -0.50 -0.94 -1.07 -1.54
ref -0.53 -0.59 -0.56 -0.82
ref -0.30 -0.50 -0.84 -1.03
ref 0.02 0.13 0.36 0.65
factors from several categories (e.g., sociocultural, economic, political, educational,
transportation, healthcare, government, religious) are necessary to promote the well-being of community
members.[19, 22±25, 38, 39, 43, 44]
Our study is distinctive in its size and scale, examining a wide range of community factors
and their relation to a comprehensive measure of well-being among a large sample of
Americans. A second strength relates to our approach. Among existing health disparities studies,
race, ethnicity, income, educational attainment, smoking status, and levels of physical activity
are all highly related to each other, and thus, their unique influence on outcomes related to
health and well-being are difficult to isolate.[
] To better identify such relationships, we
used a robust method to disentangle these community factors from each other and identify
those independently associated with well-being.
Previous county- and state-level studies have reported that a higher percent of the
population with a high school diploma was associated with higher aggregate life satisfaction or
] Our analysis extends this prior work in two major ways. First, we demonstrated
the independent association of education with a multi-dimensional assessment of well-being
at the individual level, as compared to regionally aggregated life satisfaction.[
we identified independent associations across a much broader array of community factors
including those related to clinical care, health behaviors, and the physical environment, in
addition to other social and economic factors. These results are not only more robust but also
County race composition, specifically, higher percent black, was associated with higher
iWBS and LEI scores in our final models. A prior study assessing correlates of well-being at
the state level similarly reported that the well-being index was positively associated with
measures of inclusiveness.[
] These findings are consistent with prior studies reporting that
minority status is associated with greater eudemonic well-being (i.e., an individual's judgments
about the meaning and purpose in one's life).[
] Though our findings are inconsistent
with results from one study stating that racial/ethnic minorities have lower life satisfaction,
 this negative association may be explained by higher perceived discrimination among
participants.[52±55] Taken together, the results of these studies and ours suggest that greater
tolerance may raise well-being for all community members.
Importantly, many of the county factors we identified as independently associated with well-being may be modifiable in the short to medium term and thus suitable targets for
11 / 18
Fig 1. Marginal R2 of variables in iWBS final model. Comparison of the magnitude of reduction in county level variance explained if that variable was excluded from
the model (marginal R2). Factors with larger marginal R2 contribute more to the final model than factors with a smaller marginal R2.
improving resident well-being. With two county factors related to transportation and
commuting remaining in our final model for individual well-being, and the physical environment
category itself explaining 64% of the variation in well-being, transportation and urban
planning sectors should play a key role in designing and testing interventions that may improve
the well-being of community residents. Our results clarify prior observations that long
commutes and commuting alone by automobile are correlated with low levels of positive affect and
17, 56, 57
] and suggest potential solutions. Commuting and workplace policies
such as flexibility to work from home and availability of protected bicycle lanes should be
evaluated to assess their impact on well-being of residents.
Additionally, healthcare sector variables (clinical care category), when modeled alone,
explained 80% of the variation in well-being, and several factors related to access to primary
care and screening behaviors were independently associated with well-being in the final
model. These findings extend prior observations that beyond income, perceived access to
healthcare is a strong predictor of well-being. Notably, even in countries with universal
healthcare coverage, perceived access to healthcare is correlated with well-being.[
] We show that
12 / 18
ref -0.41 -0.54 -0.56 -0.80
ref -0.37 -0.48 -0.47 -0.63
objective markers of access to primary care and utilization of preventive care are, in fact,
independently correlated with higher well-being. Though it is possible that access to primary care
is simply a marker of an unknown confounder related to higher well-being, it is also plausible
that access to healthcare mediates increased well-being by providing greater ªpeace of mindº
and necessary support for preventive health and general self-management behaviors.[
this logic holds, it implies that improving access to primary care for all community members
may enhance well-being.
Our analysis has limitations. As a cross-sectional analysis, it is not possible to make causal
inferences. Though we included a diverse set of variables in our analysis, it is possible that
associations found were due to confounding by other, unmeasured variables. With more data on
well-being gathered over time, longitudinal associations may be performed in the future to
assess the directionality of associations between well-being and county factors. Also, we relied
on secondary data sources to identify county level attributes, which meant we were unable to
capture all categories equally. For example, though we posited psychosocial variables such as
social support and tolerance could influence well-being, we were unable to include them
explicitly in our model because nationwide, county-level measures for these factors were
unavailable. Though we were limited by the available data, our selection of candidate variables
was purposeful and intended to test specific hypotheses. In addition, our well-being data were
collected during the years following the economic recession. While it is possible this may limit
the generalizability of our results since this is a particularly volatile economic period, all
noneconomic variables remaining in the final model were associated with well-being, independent
of the included contemporaneous financial and economic factors. Several of these variables
were unrelated to financial well-being, such as density of primary care physicians, along with
child poverty rate, which is directly related to financial well-being, suggesting our results may
be more broadly applicable. A final limitation is our use of county as a measure of community;
because our data are at the county level, relationships that exist at the community or city-level
could be missed. Nevertheless, our county-level results have important policy implications and
can inform local communities in developing and testing targeted programs to enhance
Our findings highlight twelve county attributes associated with higher resident well-being.
Prospective evaluation is required to assess whether changes in these county factors result in
higher well-being and new models such as the CMS accountable health communities may
facilitate this assessment.[
] Future research should evaluate whether improvements in
education, fostering diversity, the creation of healthcare delivery models that support preventive
care, and the development of environmental infrastructure that supports physical activity
14 / 18
actually improve the well-being of community members over time. Additionally, because
prior work has shown that state-level measures of inclusiveness and social tolerance are
correlated with greater well-being,[
] we encourage policymakers and public health officials to
include more psychosocial assessments in population surveys to provide necessary data for
studies that examine these factors' contribution to well-being. Future work should not only
assess the independent influence of education, healthcare, environment, and psychosocial
community factors on resident well-being, but also examine the impact of combinations of
S1 Fig. Conceptual model. Theoretical model with the initial 114 pre-specified county factors
within six categories postulated to influence various domains of resident well-being. Italicized
text denotes factors that were excluded from the study due to insufficient county level data
S1 Table. Descriptions of Gallup-Sharecare (previously Gallup-Healthways) Well-being
Index domains and survey items [
S2 Table. Mean resident life evaluation index (LEI) scores across quintiles of 77 county
factors. Each factor was categorized by equally distributed quintiles, unless noted in parentheses.
Bivariate associations for each county factor with resident well-being were tested and level of significance is noted by the Wald P-value. (DOCX)
S3 Table. Category-specific models for the life evaluation index (LEI). Correlation
coefficients for the association between each variable that was significantly associated with LEI in
bivariate analyses, independent of other variables within the same category. Each factor was
categorized by equally distributed quintiles, unless noted in parentheses. P-value reported is
the Wald P-value for trend across quintiles. R2 is the amount of variance in resident well-being
explained by all factors within each category.
We would like to thank Dr. Brent Hamar and Larissa Loufman for their valuable contributions to the final revision of this manuscript.
Conceptualization: Brita Roy, Carley Riley, Jeph Herrin, Erica S. Spatz, Elizabeth Y. Rula,
Harlan M. Krumholz.
Data curation: Brita Roy, Jeph Herrin, Kenneth P. Kell, Elizabeth Y. Rula.
Formal analysis: Jeph Herrin.
Funding acquisition: Brita Roy, Harlan M. Krumholz.
Investigation: Carley Riley.
Methodology: Brita Roy, Carley Riley, Jeph Herrin, Erica S. Spatz, Kenneth P. Kell, Elizabeth
Y. Rula, Harlan M. Krumholz.
15 / 18
Project administration: Brita Roy, John Welsh.
Resources: Harlan M. Krumholz.
Software: Jeph Herrin.
Supervision: Brita Roy, Harlan M. Krumholz.
Visualization: Brita Roy, Jeph Herrin.
Writing ± original draft: Brita Roy, Carley Riley, Jeph Herrin, Erica S. Spatz, Anita Arora,
Kenneth P. Kell, John Welsh, Elizabeth Y. Rula, Harlan M. Krumholz.
Writing ± review & editing: Brita Roy, Carley Riley, Jeph Herrin, Erica S. Spatz, Anita Arora,
Kenneth P. Kell, John Welsh, Elizabeth Y. Rula, Harlan M. Krumholz.
16 / 18
17 / 18
1. Diener E , Lucas RE , Oishi S. Subjective well-being. Handbook of positive psychology . 2002 ; 16 ( 2 ): 63 ± 73 .
2. Diener E. Subjective well-being: The science of happiness and a proposal for a national index . American psychologist . 2000 ; 55 ( 1 ): 34 . PMID: 11392863
3. Kahneman D , Diener E , Schwarz N. Well-being: Foundations of hedonic psychology: Russell Sage Foundation; 1999 .
4. Kahneman D , Riis J. Living , and thinking about it: Two perspectives on life . In: Huppert FA , Baylis N , Keverne B , editors. The Science of Well-Being: Oup Oxford; 2005 . p. 285 ± 304 .
5. Seligman ME . Flourish: A visionary new understanding of happiness and well-being: Simon and Schuster; 2012 .
6. Sears LE , Shi Y , Coberley CR , Pope JE . Overall Well-Being as a Predictor of Health Care, Productivity, and Retention Outcomes in a Large Employer . Population Health Management . 2013 ; 16 ( 6 ): 397 ± 405 . https://doi.org/10.1089/pop. 2012 .0114 PMID: 23480368
7. Rozanski A . Behavioral CardiologyCurrent Advances and Future Directions . Journal of the American College of Cardiology . 2014 ; 64 ( 1 ): 100 ± 10 . https://doi.org/10.1016/j.jacc. 2014 . 03 .047 PMID: 24998134
8. Pressman SD , Cohen S. Does Positive Affect Influence Health? Psychological Bulletin . 2005 ; 131 ( 6 ): 925 ± 71 . https://doi.org/10.1037/ 0033 - 2909 . 131 .6.925 PMID: 16351329
9. Howell RT , Kern ML , Lyubomirsky S. Health benefits: Meta-analytically determining the impact of wellbeing on objective health outcomes . Health Psychology Review . 2007 ; 1 ( 1 ): 83 ± 136 . https://doi.org/10. 1080/17437190701492486
10. Lyubomirsky S , King L , Diener E. The Benefits of Frequent Positive Affect: Does Happiness Lead to Success? Psychological Bulletin . 2005 ; 131 ( 6 ): 803 ± 55 . https://doi.org/10.1037/ 0033 - 2909 . 131 .6.803 PMID: 16351326
11. Steptoe A , Deaton A , Stone AA . Subjective wellbeing, health, and ageing . The Lancet . 2015 ; 385 ( 9968 ): 640 ± 8 .
12. Chida Y , Steptoe A. Positive Psychological Well-Being and Mortality: A Quantitative Review of Prospective Observational Studies . Psychosomatic Medicine . 2008 ; 70 ( 7 ): 741 ± 56 . https://doi.org/10.1097/ PSY.0b013e31818105ba PubMed PMID: 00006842 - 200809000 - 00001 . PMID: 18725425
13. Arora A , Spatz ES , Herrin J , Riley C , Roy B , Kell K , et al. Population Well-Being Measures Help Explain Geographic Disparities in Life Expectancy at the County Level . Health Affairs . 2016 ; 35 ( 11 ):In press.
14. Srinivasan S , O'Fallon LR , Dearry A . Creating healthy communities, healthy homes, healthy people: initiating a research agenda on the built environment and public health . American journal of public health . 2003 ; 93 ( 9 ): 1446 ± 50 . PMID: 12948961
15. Prilleltensky I , Prilleltensky O . Promoting well-being: Linking personal, organizational , and community change: John Wiley & Sons; 2007 .
16. Sampson RJ . The neighborhood context of well-being. Perspectives in biology and medicine . 2003 ; 46 ( 3 Suppl) : S53 ± 64 . Epub 2003/10/18. PMID: 14563074 .
17. Lawless NM , Lucas RE . Predictors of Regional Well-Being: A County Level Analysis . Social Indicators Research . 2011 ; 101 ( 3 ): 341 ± 57 . https://doi.org/10.1007/s11205-010-9667-7
18. Dannenberg AL , Jackson RJ , Frumkin H , Schieber RA , Pratt M , Kochtitzky C , et al. The Impact of Community Design and Land-Use Choices on Public Health: A Scientific Research Agenda . American journal of public health . 2003 ; 93 ( 9 ): 1500 ± 8 . PubMed PMID : PMC1448000 . PMID: 12948970
19. Alley DE , Asomugha CN , Conway PH , Sanghavi DM . Accountable health communitiesÐaddressing social needs through Medicare and Medicaid . The New England journal of medicine . 2016 ; 374 ( 1 ):8± 11 . https://doi.org/10.1056/NEJMp1512532 PMID: 26731305
20. Herrin J , St Andre J , Kenward K , Joshi MS , Audet AM , Hines SC . Community factors and hospital readmission rates . Health services research . 2015 ; 50 ( 1 ): 20 ± 39 . Epub 2014/04/10. https://doi.org/10.1111/ 1475 - 6773 .12177 PMID: 24712374; PubMed Central PMCID : PMCPMC4319869 .
21. Institute UoWPH. Robert Wood Johnson Foundation County Health Rankings and Roadmaps 2014 [ cited 2015 July 7] . Available from: http://www.countyhealthrankings.org/our-approach.
22. Kindig D , Stoddart G . What Is Population Health? American Journal of Public Health . 2003 ; 93 ( 3 ): 380 ± 3. https://doi.org/10.2105/AJPH.93.3.380 PMID: 12604476
23. Jacobson DM , Teutsch S . An Environmental Scan of Integrated Approaches for Defining and Measuring Total Population Health .
24. Hamilton N , Bhatti T. Population health promotion: An integrated model of population health and health promotion: Health Promotion Development Division; 1996 .
25. Van der Maesen LJ , Walker AC . Indicators of social quality: Outcomes of the European scientific network . The European Journal of Social Quality . 2005 ; 5 ( 1 /2):8± 24 .
26. Cantril H . Pattern of human concerns . 1965 .
27. Huppert FA , So TT . Flourishing across Europe: Application of a new conceptual framework for defining well-being . Social Indicators Research . 2013 ; 110 ( 3 ): 837 ± 61 . https://doi.org/10.1007/s11205-011 - 9966-7 PMID: 23329863
28. Gallup-Healthways . Gallup Daily Methodology . 2014 .
29. Diener E. Guidelines for National Indicators of Subjective Well-Being and Ill-Being . Applied Research in Quality of Life. 2006 ; 1(2):151±7 . https://doi.org/10.1007/s11482-006-9007-x
30. Diener E , Seligman ME . Beyond money toward an economy of well-being. Psychological science in the public interest . 2004 ; 5 ( 1 ):1± 31 . https://doi.org/10.1111/j.0963- 7214 . 2004 . 00501001 . x PMID : 26158992
31. Kahneman D , Riis J. Living , and thinking about it: Two perspectives on life . The science of well-being . 2005 : 285 ± 304 .
32. Diener E . Assessing subjective well-being: Progress and opportunities . Social indicators research . 1994 ; 31 ( 2 ): 103 ± 57 .
33. Gallup i. Gallup-Healthways Well-Being Index: a Methodology Report for Indexes . 2009 .
34. Evers KE , Prochaska JO , Castle PH , Johnson JL , Prochaska JM , Harrison PL , et al. Development of an individual well-being scores assessment . Psychology of Well-Being: Theory, Research and Practice . 2012 ; 2(1):1±9 . https://doi.org/10.1186/ 2211 -1522-2-2.
35. Kindig DA . http://www.improvingpopulationhealth.org/blog/2011/02/which_outcomes_improve. html2011.
36. RG P. Measuring population health outcomes . Prev Chronic Dis . 2010 .
37. Institute UoWPH. County Health Rankings & Roadmaps 2010 . Available from: http://www. countyhealthrankings.org/our-approach.
38. Bronfenbrenner U. Ecology of the family as a context for human development: Research perspectives . Developmental psychology. 1986 ; 22 ( 6 ): 723 .
39. Institute UoWPH. County Health Rankings & Roadmaps 2014 . Available from: http://www. countyhealthrankings.org/our-approach.
40. Kreft I , De Leeuw JD. Introducing Multilevel Modeling: Sage; 1998 .
Belsley DA , Kuh E , Welsch RE . Regression diagnostics . J. Wiley & Sons, New York, New York; 1980 .
42. Bradley EH , Curry LA , Spatz ES , Herrin J , Cherlin EJ , Curtis JP , et al. Hospital Strategies for Reducing Risk-Standardized Mortality Rates in Acute Myocardial Infarction . Annals of Internal Medicine . 2012 ; 156 ( 9 ): 618 ± 26 . https://doi.org/10.7326/ 0003 -4819-156-9- 201205010 -00003 PMID: 22547471
43. Derges J , Clow A , Lynch R , Jain S , Phillips G , Petticrew M , et al. ` Well London'and the benefits of participation: results of a qualitative study nested in a cluster randomised trial . BMJ open . 2014 ; 4 ( 4 ): e003596. https://doi.org/10.1136/bmjopen-2013 -003596 PMID: 24694622
44. Phillips G , Bottomley C , Schmidt E , Tobi P , Lais S , Yu G , et al. Measures of exposure to the Well London Phase-1 intervention and their association with health well-being and social outcomes . Journal of epidemiology and community health . 2014 ; 68 ( 7 ): 597 ± 605 . https://doi.org/10.1136/jech-2013 -202507 PMID: 24516117
45. Kawachi I , Daniels N , Robinson DE . Health disparities by race and class: why both matter . Health Affairs . 2005 ; 24 ( 2 ): 343 ± 52 . https://doi.org/10.1377/hlthaff.24.2.343 PMID: 15757918
46. LaVeist TA. Disentangling race and socioeconomic status: a key to understanding health inequalities . Journal of Urban Health . 2005 ; 82 ( 3 ):iii26± iii34 .
47. Rentfrow PJ , Mellander C , Florida R. Happy States of America: A state-level analysis of psychological, economic, and social well-being . Journal of Research in Personality . 2009 ; 43 ( 6 ): 1073 ± 82 . http://dx. doi.org/10.1016/j.jrp. 2009 . 08 .005.
48. Forgeard MJ , Jayawickreme E , Kern ML , Seligman ME . Doing the right thing: Measuring wellbeing for public policy . International Journal of Wellbeing . 2011 ; 1 ( 1 ).
49. Gasper D. Human Well-being: Concepts and Conceptualizations . In: McGillivray M, editor . Human Well-Being: Concept and Measurement . London: Palgrave Macmillan UK; 2007 . p. 23 ± 64 .
50. Ryff CD , Corey LMK , Hughes DL . Status Inequalities, Perceived Discrimination, and Eudaimonic WellBeing: Do the Challenges of Minority Life Hone Purpose and Growth? Journal of Health and Social Behavior . 2003 ; 44 ( 3 ): 275 ± 91 . https://doi.org/10.2307/1519779 PMID: 14582308
51. Barger SD , Donoho CJ , Wayment HA . The relative contributions of race/ethnicity, socioeconomic status, health, and social relationships to life satisfaction in the United States . Quality of Life Research . 2009 ; 18 ( 2 ): 179 ± 89 . https://doi.org/10.1007/s11136-008 -9426-2 PMID: 19082871
52. Barnes PW , Lightsey OR . Perceived racist discrimination, coping, stress, and life satisfaction . Journal of Multicultural Counseling and Development . 2005 ; 33 ( 1 ): 48 ± 61 .
53. Branscombe NR , Schmitt MT , Harvey RD . Perceiving pervasive discrimination among African Americans: Implications for group identification and well-being . Journal of Personality and Social Psychology . 1999 ; 77 ( 1 ): 135 ± 49 . https://doi.org/10.1037/ 0022 - 3514 . 77 .1. 135
54. Broman CL . Race-related factors and life satisfaction among African Americans . Journal of Black Psychology . 1997 ; 23 ( 1 ): 36 ± 49 .
55. Verkuyten M. Life satisfaction among ethnic minorities: The role of discrimination and group identification . Social indicators research . 2008 ; 89 ( 3 ): 391 ± 404 .
56. Kahneman D , Krueger AB , Schkade D , Schwarz N , Stone A. Toward National Well-Being Accounts . The American Economic Review . 2004 ; 94 ( 2 ): 429 ± 34 .
57. Diener E , Tov W. National Accounts of Well-Being . In: Land CK , Michalos CA , Sirgy JM , editors. Handbook of Social Indicators and Quality of Life Research . Dordrecht: Springer Netherlands; 2012 . p. 137 ± 57 .
58. Jorgensen BS , Jamieson RD , Martin JF . Income, sense of community and subjective well-being: Combining economic and psychological variables . Journal of Economic Psychology . 2010 ; 31 ( 4 ): 612 ± 23 . http://dx.doi.org/10.1016/j.joep. 2010 . 04 .002.
59. Patrick H , Williams GC . Self-determination theory: its application to health behavior and complementarity with motivational interviewing . International Journal of behavioral nutrition and physical Activity . 2012 ; 9 ( 1 ): 1 .