Geodemographic Disparities in Availability of Comprehensive Intimate Partner Violence Screening Services in Miami-Dade County,Florida

Abstract

This study examined geodemographic factors associated with availability of comprehensive intimate partner violence (IPV) screening services in Miami-Dade County, Florida. We geocoded 2014 survey data from 278 health facilities and created a population-normalized density surface of IPV screening comprehensiveness. We used correlation analysis and spatial regression techniques to evaluate census tract-level predictors of the mean normalized comprehensiveness score (NCS) for 505 census tracts in Miami-Dade. The population-adjusted density surface of IPV screening comprehensiveness revealed geographic disparities in the availability of screening services. Using a spatial lag regression model, we observed that race and ethnicity are associated with mean NCS by census tract after controlling for age, median gross rent, and receipt of Social Security benefits. The percentage of White non-Hispanic residents was positively associated with NCS, Black non-Hispanic was negatively associated with NCS, while Hispanic—the majority ethnicity in Miami-Dade—was not associated with NCS. This exploratory study may be the first to put IPV screening comprehensiveness on the map, and provides a starting point for addressing urban disparities in the availability of IPV screening services that are shaped by race, ethnicity, zoning, and socioeconomic status.

Keywords

intimate partner violence IPV screening geographic information systems health disparities

Introduction

Intimate partner violence (IPV) comprises actions or threats of physical, sexual, psychological, or emotional abuse by a current or former spouse or partner (Saltzman, Fanslow, McMahon, & Shelley, 2002). IPV is a significant public health burden, as about a third of U.S. women report experiencing physical violence from an intimate partner in their lifetime, and women who endure IPV are more likely to experience worse health outcomes (Breiding, Chen, & Black, 2014). Because victims of IPV are more likely to utilize health care services relative to nonvictims (Breiding et al., 2014; Ulrich et al., 2003), health care facilities present an opportunity for surveillance and intervention of IPV. Despite evidence that IPV detection and early intervention result in numerous health benefits with minimal risks (Houry et al., 2008; Nelson, Bougatsos, & Blazina, 2012), little is known about how IPV care is implemented in practice.

Several studies have found that screening for IPV is low, ranging from 1.5% to 18%, and varies by the type of health care provider (i.e., primary care, obstetricians, etc.) (Borowsky & Ireland, 2002; Rodriguez, Bauer, McLoughlin, & Grumbach, 1999; Waalen, Goodwin, Spitz, Petersen, & Saltzman, 2000). IPV help-seeking behavior is complex and has been linked to race and ethnicity (Ingram, 2007; Lipsky, Caetano, Field, & Larkin, 2006), informal support structures (Barrett & Pierre, 2011), and balancing readiness for help with access constraints (Robinson & Spilsbury, 2008), with low acculturation decreasing utilization by Hispanic women victims (Lipsky et al., 2006). A variety of provider- and patient-related barriers, such as personal discomfort discussing IPV, and lack of knowledge or time, continue to impede widespread implementation of IPV screening (Sprague et al., 2012). The comprehensiveness of a screening program has been shown to improve effectiveness in identifying IPV (O’Campo, Kirst, Tsamis, Chambers, & Ahmad, 2011). A recent study of IPV screening and response policies among health care facilities in Miami-Dade County, Florida, revealed wide variation in the characteristics and comprehensiveness of IPV care, as well as gaps in our knowledge of patient outcomes (Williams, Halstead, Salani, & Koermer, 2016). It is also unclear to what extent the increase in IPV advocacy and care over the last decade has been able to address geodemographic disparities in service availability and access.

The sociodemographic risk factors for IPV have been well-studied (Breiding, Black, & Ryan, 2008; Capaldi, Knoble, Shortt, & Kim, 2012; García-Moreno, Jansen, Ellsberg, Heise, & Watts, 2005), with U.S. studies tending to emphasize the role of poverty and alcohol consumption. The geodemographic factors associated with IPV have not been extensively studied, and have tended to focus on neighborhood characteristics associated with IPV prevalence, rather than access to IPV services. Neighborhood-level factors associated with IPV have often been framed by social disorganization theory (Pinchevsky & Wright, 2012) and have included alcohol outlet density (Cunradi, Mair, Ponicki, & Remer, 2011; Livingston, 2011; Waller et al., 2011), ethnic heterogeneity (Wright & Benson, 2010), and residential stability (Li et al., 2010). In a recent review of neighborhood environmental factors associated with IPV, built environmental factors and access to services were noted as significant IPV research gaps (Beyer, Wallis, & Hamberger, 2015). A geographically framed study of Duval County, Florida, demonstrated that IPV was spatially concentrated in areas of high poverty (Miles-Doan & Kelly, 1997), while more recent studies in Valencia, Spain, observed excess IPV risk in physically disordered and decaying neighborhoods (Gracia, López-Quílez, Marco, Lladosa, & Lila, 2014, 2015). A study in Iowa revealed rural–urban disparities in geographic access to IPV resources as measured by driving distance: women in rural areas lived 3 times farther from the closest IPV resource relative to women in urban areas or large rural towns (Peek-Asa et al., 2011). Our literature review did not reveal any similar studies assessing intra-urban disparities in IPV resource availability or access.

Miami-Dade County, with over 2.6 million residents, is the seventh largest U.S. county by population (U.S. Census Bureau, 2016). The county’s geographic location in South Florida and its diverse Hispanic majority population have earned Miami-Dade a reputation as the “Gateway of the Americas.” Miami-Dade has a legacy of displacing the Black non-Hispanic (BNH) population into planned communities to accommodate urban development projects, while the wealthier White non-Hispanic (WNH) population has tended to settle along coastal and beach areas (Nijman, 2010). The resulting ethnic segregation of Miami-Dade yields distinct geographic patterns for BNH, WNH, and Hispanic residents, which together comprise 97% of Miami-Dade’s population (Figure 1) (U.S. Census Bureau, 2016). Miami-Dade’s ethnic distribution interacts with zoning patterns to produce BNH and WNH enclaves that often result in similarly low geographic availability of health care facilities, often driven by substantial inequalities in economic access. Many WNH enclaves tend to be wealthier neighborhoods that are zoned homogenously as residential with larger lots; while this creates longer travel distances to commercial areas, residents of these enclaves do not generally suffer constrained health care access given their socioeconomic standing. Many BNH enclaves are relatively higher density areas with longer travel distances to commercial areas due to lower car ownership, historical municipal disinvestment, lower property values, and thus less community capacity to attract higher order commercial tenants such as health care providers. Geography and race are well-known drivers of access to health services (Chandra & Skinner, 2004; McLafferty, 2003), and these disadvantages are compounded by generally lower levels of education and socioeconomic status among Miami-Dade’s BNH community, which reinforce residential segregation (Boswell & Cruz-Báez, 1997).

Figure 1.

Smoothed population density surfaces for (a) Hispanic, (b) White non-Hispanic, and (c) Black non-Hispanic residents generated from 2010 census block population counts of Miami-Dade County, FL.

Using data from a survey of IPV screening comprehensiveness among Miami-Dade health care facilities, we perform an exploratory ecological analysis of the correlates of IPV screening availability. We operationalize IPV screening availability via a census tract-level comprehensiveness score for IPV screening services that is normalized by total population. We first compute and map this “normalized comprehensiveness score” (NCS) to understand variation in availability across Miami-Dade. We then hypothesize that the percentage of BNH population in a census tract will be negatively associated with NCS, while the percentage of WNH will be positively associated with NCS, after controlling for sociodemographic factors.

Method

Data

The health care facility survey used for this study was designed to assess variation in IPV screening and response policies and procedures across Miami-Dade; full details are available elsewhere (Williams et al., 2016). Between June 2014 and January 2015, phone surveys were completed with 288 randomly selected primary care, obstetrics/gynecology, pediatric, and emergency department facilities out of a county-wide sampling frame of 1,208. Facilities were given a binary score (0 or 1) for six dimensions assessing the comprehensiveness of IPV screening and response: (a) routine screening, (b) consistent screening measure, (c) referral/response procedures, (d) provider/staff training, (e) fidelity monitoring, and (f) written policies. The sum of these responses is the comprehensiveness score, which is further classified as an ordinal measure to minimize bias in the responses of some facility interviewees (Williams et al., 2016). Facilities that reported five or six screening components were classified as high comprehensiveness, three or four as medium comprehensiveness, one or two as low comprehensiveness, and zero as no comprehensiveness. All procedures were reviewed by the University of Miami Institutional Review Board and deemed to be nonhuman subjects research. This analysis was completed in early 2016.

Measures

NCS

We used a geographic information system (GIS) to geocode all 1,208 facilities in the original health facility sampling frame and used the Moran’s I statistic to test for any spatial autocorrelation among the 288 surveyed facilities and the four health facility types (Figure 2). We then created a Gaussian kernel density surface from the IPV comprehensiveness score of 278 surveyed facilities (ten facilities had incomplete survey data and no comprehensiveness score), with high, medium, low, and no comprehensiveness represented as 3, 2, 1, and 0. The use of kernel density layers to estimate spatial accessibility is well established as an alternative to traditional gravity models and floating catchment area methods (Guagliardo, 2004). We experimented with several kernel sizes and ultimately specified a 1 km kernel, a geographic window that is equivalent to a 10- to 15-min walk and maximized variability in the resulting surface. We applied an inverse-distance weighting regime, which means that a pixel’s score density value is inversely related to its distance from nearby facilities. The resulting pixel resolution was 515 feet (see Figure 3a). We created a population density surface from the centroids of 2010 census block population counts using the same parameters so that the resulting surface has the exact same pixel resolution as the comprehensiveness score density layer (Figure 3b). We implemented simple map algebra to divide the comprehensiveness score density layer by the population density layer, and the resulting layer was rescaled to represent the density of comprehensiveness scores per 10,000 residents, that is, the NCS (Figure 3c). Finally we summarize the mean NCS pixel value by census tract to explore correlates of NCS with aggregated U.S. Census data; this mean NCS value is the dependent measure.

Figure 2.

(a) The sampling frame and facility participants in the IPV screening comprehensiveness survey, and (b) the distribution of health care facility types.

Figure 3.

Kernel density surfaces for (a) comprehensiveness score, with geocoded facilities and their respective IPV screening comprehensiveness scores; (b) total population for Miami-Dade; and (c) the normalized comprehensiveness score (NCS), the quotient of comprehensiveness score density divided by population density.

Census characteristics

We explore correlates of NCS by modeling mean NCS against a list of independent measures aggregated by census tract from the U.S. Census decennial product (2010) and U.S. Census rolling 5-year American Community Survey product (2006-2010). We selected typical demographic and socioeconomic characteristics related to poverty, race, age, education, and neighborhood characterization that are traditionally associated with disparities in access to health services (Bissonnette, Wilson, Bell, & Shah, 2012; Fiscella, Franks, Gold, & Clancy, 2000) and IPV prevalence (Cunradi, Caetano, & Schafer, 2002; Rennison & Planty, 2003) (see Table 1).

Table 1.

2010 Census Tract Characteristics for Miami-Dade County (N = 505), and Pearson’s Correlations (r) With the NCS.

Characteristic	Tract M (SD)	Pearson’s Correlation (r) With NCS
Black non-Hispanic (%)	18.69 (26.66)	−.044
White non-Hispanic (%)	17.43 (17.38)	.102*
Hispanic (%)	63.13 (27.02)	−.031
Median age (years)	38.77 (5.90)	.061
No high school diploma (%)	21.70 (13.55)	−.010
Limited English proficiency (%)	32.75 (18.10)	.001
Median gross rent (US$)	1,173.27 (371.74)	−.120**
Per capita income (US$)	23,574.17 (14,902.88)	.089*
Renter-occupied housing units (%)	38.70 (20.86)	.091*
Population below poverty line (%)	17.23 (12.42)	.012
Receives social security benefits (%)	26.56 (11.32)	.125**
Female-headed households (%)	18.93 (9.07)	−.059
Mobile home housing units (%)	1.89 (8.81)	−.088*
Housing units with no automobile (%)	11.34 (11.80)	.042

Note. NCS = normalized comprehensiveness score.

p < .05. **p < .01.

Statistical Analysis

We performed bivariate correlation and multivariate ordinary least squares (OLS) regression analyses using SPSS 22 (IBM, Armonk, NY). We used stepwise procedures as an additional exploratory tool to guide the model building process by highlighting variables with the strongest association with NCS in a multivariate context. We strived for the most parsimonious model, that is, the model with the strongest explanatory power with as few independent terms as possible. Spatial data were managed and analyzed using ArcGIS 10.3.1 (ESRI, Redlands, CA). Because spatial dependence was imputed into the dependent NCS measure via the smoothing effect of the kernel density procedures, we confirmed any spatial autocorrelation in the residuals of candidate models using the Moran’s I statistic. We ultimately control for spatial autocorrelation by fitting a spatial lag regression model using GeoDa 1.8.2 (Arizona State University, Tempe, AZ) of the basic form in Equation 1:

y = ρ W y + Χ β + ε,

where W is the spatial weights matrix (neighbors are defined by first-order contiguity among census tracts), $ρ$ measures the degree of spatial autocorrelation (between −1 and 1), and thus $ρ W y$ measures the spatial dependence in $y$ . Correlation and regression analyses are based on 505 census tracts that fall within the Miami-Dade County-defined urban development boundary, and are interpreted using a significance threshold of α = .05.

Results

The geocoded facilities and their respective IPV screening comprehensiveness scores are presented, along with the density surface for comprehensiveness score, in Figure 3a. The 278 surveyed facilities were first assessed for spatial autocorrelation using the Moran’s I statistic to determine whether the original (aspatial) survey design inadvertently resulted in any spatial biases. We observed no evidence of statistically significant spatial patterning among the 278 surveyed facilities relative to the sampling frame, among the distribution of comprehensiveness scores, or among each facility type relative to all other types.

The population density surface for Miami-Dade is presented in Figure 3b, and the resulting NCS surface is in Figure 3c. All three maps in Figure 3 use a standard deviation classification scheme to show the most extreme values in dark tones. The NCS surface provides a standardized metric of IPV screening comprehensiveness per 10,000 residents that facilitates comparison between areas within Miami-Dade. NCS ranged from 0 to 12.68, with median of 1.41, and mean of 1.90 (SD = 1.87). The largest region of high NCS is located southwest of downtown Miami between Coral Gables and Kendall, but focused along the commercial corridors of Kendall Drive and South Dixie Highway (US 1). Additional regions of high NCS are located to the south in Palmetto Bay, to the north near Miami Gardens, and in central Miami-Dade in Hialeah and Doral. Pockets of low NCS exist in the predominantly BNH neighborhoods of Opa-locka, South Miami Heights, and Florida City, as well as some smaller suburban communities in southern Miami-Dade.

Table 1 presents bivariate Pearson’s correlation coefficients (r) for 14 candidate predictors of NCS; only six measures were significantly correlated with NCS. Median gross rent (r = −.120, p = .007) and the percent of mobile home housing units (r = −.088, p = .047) were both significantly negatively correlated with NCS. The percent receiving Social Security benefits (r = .125, p = .005), WNH population (r = .102, p = .022), renter-occupied housing units (r = .091, p = .042), and per capita income (r = .089, p = .045) were all significantly positively correlated with NCS. The negative correlation between BNH and NCS was not statistically significant (r = −.044, p = .326).

We iteratively fit a series of multivariate OLS regression models of NCS guided by the results in Table 1 and by the results of forward and backward stepwise regression models. All of these models revealed statistically significant positive spatial autocorrelation in their residuals, so we could not properly specify an OLS model. The Lagrange Multiplier tests for our candidate models were significant and indicated that a spatial lag model was a better fit. Table 2 presents the results of the spatial lag model of NCS that best minimized the Akaike information criterion (AIC = 1409.11, R² = .79) with the fewest number of covariates. The spatial lag term, $ρ$ , has the strongest association with NCS (β = .93, z = 61.69, p < .001), because it is recapturing and controlling for much of the spatial dependence in NCS created by the kernel density procedure. Additional model diagnostics indicate that there is still some unresolved spatial dependence.

Table 2.

Spatial Lag Regression Model of the Normalized Comprehensiveness Score on Select Sociodemographic Characteristics for 505 Census Tracts in Miami-Dade County.

Characteristic	β (SE)	z Score	p Value
Constant	1.39 (.38)	3.62	<.001
Spatial lag term (Rho)	0.93 (.02)	61.69	<.001
Median age (years)	−0.03 (.01)	−2.89	.004
Median gross rent (US$)	−0.00 (.00)	−2.77	.006
Black non-Hispanic (%)	−0.35 (.18)	−1.90	.057
White non-Hispanic (%)	0.58 (.26)	2.22	.026
Receives social security benefits (%)	0.01 (.01)	2.24	.025
Model diagnostics: AIC = 1409.11, R² = .79

Note. AIC = Akaike information criterion.

Median age (β = −.03, z = −2.89, p = .004) and median gross rent (β = −.00, z = −2.77, p = .006) were both negatively associated with NCS. This probably reflects zoning differences, as younger, lower income residents are more likely to live in commercial- or mixed-zoning areas that accommodate health care facilities. Receiving Social Security benefits was positively associated with NCS (β = .01, z = 2.24, p = .025), which is consistent with zoning for lower income residents. But because Social Security benefit utilization is generally a proxy for older adults, this finding contradicts the negative association between NCS and median age. Finally, our hypothesis about race and ethnicity is partially supported, as WNH is positively associated with NCS (β = .58, z = 2.22, p = .026), while BNH is negatively associated with NCS (β = −.35, z = −1.90, p = .057), though the finding was not statistically significant.

Discussion

This study mapped IPV screening comprehensiveness scores for 278 health care facilities in Miami-Dade County, created a population-normalized surface of screening comprehensiveness density, and explored geodemographic factors associated with screening comprehensiveness by census tract. To our knowledge, this is the first exploratory study of intra-urban availability of IPV screening resources in a large metropolitan area. We find that race and ethnicity are associated with mean NCS by census tract after controlling for age, median gross rent, and receipt of Social Security benefits. WNH was positively associated with NCS, BNH was negatively associated with NCS, while Hispanic—the majority ethnicity in Miami-Dade—was not associated with NCS.

Our hypothesis was only partially supported, as the relationship between BNH and NCS was nonsignificant, though it approached statistical significance. Although the concentration of BNH residents between downtown Miami and Miami Gardens generally appears underserved, several BNH-majority census tracts around the county have a high mean NCS. These include tracts in southeast Miami Gardens (a commercial strip at the convergence of the Florida Turnpike and I-95 Highway), downtown Miami (next to the Jackson Memorial Hospital campus), and BNH enclaves neighboring Coral Gables and Palmetto Bay. Although these neighborhoods may temper the association between BNH and the availability of comprehensive IPV screening, we would still expect differences in access to screening services given the stark socioeconomic differences between BNH- and WNH-majority tracts. In the 80 tracts that are at least 50% BNH, median gross rent is US$907 and the median house value is US$198,000; in the 42 tracts that are at least 50% WNH, median gross rent is US$1,442 and the median house value is US$559,000.

Although there are no known studies of IPV screening geodemographics for direct comparison, these results are consistent, in part, with a recent review of IPV risk factors in which younger age, deprivation (including unemployment and low income), and minority group membership were predictive of IPV, with evidence of mediation by income (Capaldi et al., 2012). The same review also found that being a U.S.-born Hispanic was a risk factor relative to being foreign born, but degree of acculturation was not predictive and therefore did not explain this association. This phenomenon has been observed in other immigrant populations; neighborhoods with higher concentrations of immigrants were found to have lower levels of IPV in Chicago, a relationship mediated by cultural norms and social ties (Wright & Benson, 2010). But this issue is likely complicated by the diversity of Hispanic demographics in Miami-Dade. There is likely additional spatial heterogeneity of IPV service availability and access within Miami-Dade’s Hispanic population that cannot be assessed using the generic Hispanic label employed by the U.S. Census Bureau.

There is some apparent contradiction among covariates in our model. Tracts with the highest NCS are more WNH, less BNH, younger, and have lower median rents, yet also have more residents receiving Social Security benefits. The positive association with Social Security benefits may be an artifact of modeling zoning and sociodemographics simultaneously, particularly after controlling for age. Median gross rent is our closest proxy for zoning, and we would expect modest distance decay in rents with residential proximity to noisy commercial hubs across Miami-Dade’s suburban sprawl, zones that appeal more to young singles than families. Social Security recipients may be a proxy for both age and/or socioeconomic status given the number of Disability Insurance beneficiaries or Supplementary Security Income recipients in a census tract; it may also indicate a high number of multigenerational households. But the big picture is that after controlling for these covariates, BNH was still negatively associated with NCS and borderline statistically significant. With better controls for zoning, we might observe even starker disparities in NCS between WNH and BNH.

A key strength of this study is the uniqueness and strong sample size of the original IPV comprehensiveness survey data set, which turned out to also be geographically representative of Miami-Dade’s health care facility landscape. The sample is also a limitation, as there are many health care facilities offering IPV services that are not reflected in the NCS surface. In particular, there are a few major IPV resource centers in Miami-Dade that are not based in health care facilities: Dade County Domestic Violence in downtown Miami; Coordinated Victims Assistance Center (CVAC) in Coconut Grove; Jefferson Reaves House in Brownsville; Men and Women United in Justice, Education, and Reform (MUJER) in Homestead; and others. Additional IPV services are occasionally provided through mental health and marriage counseling centers, but the problem is that there is no comprehensive database of IPV service providers. Creating such a public clearinghouse of IPV resources is an opportunity for an immediate public health contribution that would facilitate victim assistance-seeking and clinical referrals.

The spatial lag model implemented in this study only partly controls for the spatial dependence in NCS created by the smoothing effect of the kernel density method, and any additional spatial dependence that is not captured by the chosen covariates. A recent attempt to model IPV risk in Valencia, Spain, produced estimates with additional spatial structure that was not explained by the covariates (Gracia et al., 2015), so there appear to be geographic components of both IPV risk and screening services that are difficult to measure. The unmeasured spatial component of IPV screening comprehensiveness may also be related to findings in other recent studies in which screening women was not associated with better outcomes, such as increased use of IPV support services or cessation of violence (Klevens, Sadowski, Kee, Garcia, & Lokey, 2015; MacMillan et al., 2009; O’Doherty et al., 2014). Geographic frameworks thus offer additional possibilities for studying IPV and screening services. Additional qualitative tools such as concept mapping, which has been used successfully to map neighborhood domains that elucidate pathways to IPV (O’Campo, Burke, Peak, McDonnell, & Gielen, 2005), can also be adapted to explore social and environmental barriers to IPV screening and service utilization.

Conclusion

This study presents an exploratory attempt to put IPV screening comprehensiveness on the map. The integration of IPV screening into health care settings has accelerated nationally, spurred by new recommendations and guidelines for screening and treatment by the Institute of Medicine and the U.S. Preventive Services Task Force (USPSTF) (Institute of Medicine, 2011; Moyer, 2013), and by provisions in the Affordable Care Act that allow for the reimbursement of IPV preventive services (U.S. Department of Health and Human Services [DHHS], 2013). By visualizing population-adjusted service availability, public health officials and IPV advocates can begin evaluating the implications of neighborhood-level disparities in comprehensive screening. The logical extension of this work is to use crime data as a proxy for IPV activity and assess NCS coverage relative to IPV incidence. Miami-Dade’s patchwork of incorporated cities and cultural variation in IPV reporting are logistical challenges, but this approach presents a significant opportunity to identify and target underserved areas for IPV screening and capacity building.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded in part through a University of Miami Provost Research Award (J.R.W.).

Author Biographies

Justin Stoler is an assistant professor of geography and regional studies, and public health sciences, at the University of Miami. His research explores the geographic patterns of urban health disparities and environmental influences on social and behavioral epidemiology.

Jaclyn Verity is an undergraduate student double-majoring in health geographics and religious studies at the University of Miami.

Jessica R. Williams is an assistant professor at the University of Miami School of Nursing and Health Studies. Her research aims at improving methods for the dissemination and implementation of evidence-based practices, with focus on gender-based violence prevention.

References

Barrett

B. J.

Pierre

M. S.

(2011). Variations in women’s help seeking in response to intimate partner violence: Findings from a Canadian population-based study. Violence Against Women, 17, 47-70.

Beyer

Wallis

A. B.

Hamberger

L. K.

(2015). Neighborhood environment and intimate partner violence: A systematic review. Trauma, Violence, & Abuse, 16, 16-47.

Bissonnette

Wilson

Bell

Shah

T. I.

(2012). Neighbourhoods and potential access to health care: The role of spatial and aspatial factors. Health & Place, 18, 841-853.

Borowsky

I. W.

Ireland

(2002). Parental screening for intimate partner violence by pediatricians and family physicians. Pediatrics, 110, 509-516.

Boswell

T. D.

Cruz-Báez

A. D.

(1997). Residential segregation by socioeconomic class in metropolitan Miami: 1990. Urban Geography, 18, 474-496.

Breiding

M. J.

Black

M. C.

Ryan

G. W.

(2008). Prevalence and risk factors of intimate partner violence in eighteen U.S. states/territories, 2005. American Journal of Preventive Medicine, 34, 112-118.

Breiding

M. J.

Chen

Black

(2014). Intimate partner violence in the United States—2010. Atlanta, GA: National Center for Injury Prevention and Control, Centers for Disease Control and Prevention.

Capaldi

D. M.

Knoble

N. B.

Shortt

J. W.

Kim

H. K.

(2012). A systematic review of risk factors for intimate partner violence. Partner Abuse, 3, 231-280.

Chandra

Skinner

J. S.

(2004). Geography and racial health disparities. In Anderson

N. B.

Bulatao

R. A.

Cohen

(Eds.), Critical perspectives on racial and ethnic differences in health in late life (pp. 604-640). Washington, DC: National Academies Press.

10.

Cunradi

C. B.

Caetano

Schafer

(2002). Socioeconomic predictors of intimate partner violence among White, Black, and Hispanic couples in the United States. Journal of Family Violence, 17, 377-389.

11.

Cunradi

C. B.

Mair

Ponicki

Remer

(2011). Alcohol outlets, neighborhood characteristics, and intimate partner violence: Ecological analysis of a California city. Journal of Urban Health, 88, 191-200.

12.

Fiscella

Franks

Gold

M. R.

Clancy

C. M.

(2000). Inequality in quality: Addressing socioeconomic, racial, and ethnic disparities in health care. Journal of the American Medical Association, 283, 2579-2584.

13.

García-Moreno

Jansen

H. A. F. M.

Ellsberg

Heise

Watts

(2005). WHO multi-country study on women’s health and domestic violence against women: Initial results on prevalence, health outcomes and women’s responses. Geneva, Switzerland: World Health Organization.

14.

Gracia

López-Quílez

Marco

Lladosa

Lila

(2014). Exploring neighborhood influences on small-area variations in intimate partner violence risk: A Bayesian random-effects modeling approach. International Journal of Environmental Research and Public Health, 11, 866-882.

15.

Gracia

López-Quílez

Marco

Lladosa

Lila

(2015). The spatial epidemiology of intimate partner violence: Do neighborhoods matter? American Journal of Epidemiology, 182, 58-66.

16.

Guagliardo

M. F.

(2004). Spatial accessibility of primary care: Concepts, methods and challenges. International Journal of Health Geographics, 3, 1-13.

17.

Houry

Kaslow

N. J.

Kemball

R. S.

McNutt

L. A.

Cerulli

Straus

. . . Rhodes

K. V.

(2008). Does screening in the emergency department hurt or help victims of intimate partner violence? Annals of Emergency Medicine, 51, 433-442.

18.

Ingram

E. M.

(2007). A comparison of help seeking between Latino and non-Latino victims of intimate partner violence. Violence Against Women, 13, 159-171.

19.

Institute of Medicine. (2011). Clinical preventive services for women: Closing the gaps. Washington, DC: National Academies Press.

20.

Klevens

Sadowski

L. S.

Kee

Garcia

Lokey

(2015). Effect of screening for partner violence on use of health services at 3-year follow-up of a randomized clinical trial. Journal of the American Medical Association, 314, 515-516.

21.

Kirby

R. S.

Sigler

R. T.

Hwang

S.-S.

LaGory

M. E.

Goldenberg

R. L.

(2010). A multilevel analysis of individual, household, and neighborhood correlates of intimate partner violence among low-income pregnant women in Jefferson County, Alabama. American Journal of Public Health, 100, 531-539.

22.

Lipsky

Caetano

Field

C. A.

Larkin

G. L.

(2006). The role of intimate partner violence, race, and ethnicity in help-seeking behaviors. Ethnicity & Health, 11, 81-100.

23.

Livingston

(2011). A longitudinal analysis of alcohol outlet density and domestic violence. Addiction, 106, 919-925.

24.

MacMillan

H. L.

Wathen

Jamieson

Boyle

M. H.

Shannon

H. S.

Ford-Gilboe

. . . McNutt

L. A.

(2009). Screening for intimate partner violence in health care settings: A randomized trial. Journal of the American Medical Association, 302, 493-501.

25.

McLafferty

S. L.

(2003). GIS and health care. Annual review of Public Health, 24, 25-42.

26.

Miles-Doan

Kelly

(1997). Geographic concentration of violence between intimate partners. Public Health Reports, 112, 135-141.

27.

Moyer

V. A.

(2013). Screening for intimate partner violence and abuse of elderly and vulnerable adults: U.S. Preventive Services Task Force recommendation statement. Annals of Internal Medicine, 158, 478-486.

28.

Nelson

H. D.

Bougatsos

Blazina

(2012). Screening women for intimate partner violence: A systematic review to update the U.S. Preventive Services Task Force recommendation. Annals of Internal Medicine, 156, 796-808.

29.

Nijman

(2010). Miami: Mistress of the Americas. Philadelphia: University of Pennsylvania Press.

30.

O’Campo

Burke

Peak

G. L.

McDonnell

K. A.

Gielen

A. C.

(2005). Uncovering neighbourhood influences on intimate partner violence using concept mapping. Journal of Epidemiology and Community Health, 59, 603-608.

31.

O’Campo

Kirst

Tsamis

Chambers

Ahmad

(2011). Implementing successful intimate partner violence screening programs in health care settings: Evidence generated from a realist-informed systematic review. Social Science & Medicine, 72, 855-866.

32.

O’Doherty

L. J.

Taft

Hegarty

Ramsay

Davidson

L. L.

Feder

(2014). Screening women for intimate partner violence in healthcare settings: Abridged Cochrane systematic review and meta-analysis. British Medical Journal, 348, g2913.

33.

Peek-Asa

Wallis

Harland

Beyer

Dickey

Saftlas

(2011). Rural disparity in domestic violence prevalence and access to resources. Journal of Women’s Health, 20, 1743-1749.

34.

Pinchevsky

G. M.

Wright

E. M.

(2012). The impact of neighborhoods on intimate partner violence and victimization. Trauma, Violence, & Abuse, 13, 112-132.

35.

Rennison

Planty

(2003). Nonlethal intimate partner violence: Examining race, gender, and income patterns. Violence and Victims, 18, 433-443.

36.

Robinson

Spilsbury

(2008). Systematic review of the perceptions and experiences of accessing health services by adult victims of domestic violence. Health & Social Care in the Community, 16, 16-30.

37.

Rodriguez

M. A.

Bauer

H. M.

McLoughlin

Grumbach

(1999). Screening and intervention for intimate partner abuse: Practices and attitudes of primary care physicians. Journal of the American Medical Association, 282, 468-474.

38.

Saltzman

L. E.

Fanslow

J. L.

McMahon

P. M.

Shelley

G. A.

(2002). Intimate partner violence surveillance: Uniform definitions and recommended data elements: Version 1.0. Atlanta, GA: Centers for Disease Control and Prevention, National Center for Injury Prevention and Control.

39.

Sprague

Madden

Simunovic

Godin

Pham

N. K.

Bhandari

Goslings

J. C.

(2012). Barriers to screening for intimate partner violence. Women & Health, 52, 587-605.

40.

Ulrich

Y. C.

Cain

K. C.

Sugg

N. K.

Rivara

F. P.

Rubanowice

D. M.

Thompson

R. S.

(2003). Medical care utilization patterns in women with diagnosed domestic violence. American Journal of Preventive Medicine, 24, 9-15.

41.

U.S. Census Bureau. (2016). Miami-Dade County, Florida: QuickFacts. Retrieved from http://www.census.gov/quickfacts/table/PST045214/12086

42.

U.S. Department of Health and Human Services. (2013). Affordable Care Act rules on expanding access to preventive service for women. Retrieved from www.hhs.gov/healthcare/facts/factsheets/2011/08/womensprevention08012011a.html

43.

Waalen

Goodwin

M. M.

Spitz

A. M.

Petersen

Saltzman

L. E.

(2000). Screening for intimate partner violence by health care providers: Barriers and interventions. American Journal of Preventive Medicine, 19, 230-237.

44.

Waller

M. W.

Iritani

B. J.

Christ

S. L.

Clark

H. K.

Moracco

K. E.

Halpern

C. T.

Flewelling

R. L.

(2011). Relationships among alcohol outlet density, alcohol use, and intimate partner violence victimization among young women in the United States. Journal of Interpersonal Violence, 27, 2062-2086.

45.

Williams

J. R.

Halstead

Salani

Koermer

(2016). Intimate partner violence screening and response: Policies and procedures across health care facilities. Women’s Health Issues, 26, 377-383.

46.

Wright

E. M.

Benson

M. L.

(2010). Immigration and intimate partner violence: Exploring the immigrant paradox. Social Problems, 57, 480-503.