An Examination of Geographic Differences in Social Determinants of Health Among Women with Diagnosed HIV in the United States and Puerto Rico, 2017

Data reported in the National HIV Surveillance System (NHSS) of the Centers for Disease Control and Prevention were used to determine census tract-level HIV diagnosis rates and percentages among adult women (aged ≥18 years) in 2017. Data from the American Community Survey were combined with NHSS data to examine regional differences in federal poverty status, education level, income level, employment status, and health insurance coverage among adult women with diagnosed HIV infection in the United States and Puerto Rico.


INTRODUCTION
Although the rates of new Human Immunodeficiency Virus (HIV) infections in the United States have declined greatly, women are still affected heavily, especially Black/ African American (here inafter referred to as Black) women and women living in the South [1]. HIV among Black women has decreased [1], but the rate among Black women is on ave-

METHODS
Data from CDC's National HIV Surveillance System (NHSS) and the U.S. Census Bureau's American Community Survey (ACS) were used for this analysis.

NHSS
Numbers, rates, and percentages of diagnoses of HIV infection for 2017 were based on cases reported to NHSS from 2017 to December 2018. Those include cases among women aged 18 years and older whose HIV infection was diagnosed during 2017, regardless of the stage of disease at diagnosis. HIV surveillance data were geocoded to the census tract level for addresses of residence at the time of diagnosis and linked at the census tract level to the SDH indicator variable data from the ACS.
The NHSS is a population-based database of all persons in the United States and Puerto Rico with diagnosed HIV infection, based on standard, robust statistical methodology. Annual evaluation of NHSS data show that as of December 2017, the completeness of reporting of HIV infection was estimated to be at least 85% [1]. Because NHSS is a comprehensive national surveillance system that contains data on nearly all people with diagnosed HIV in the United States, no statistical tests of significance were applied to the data.

ACS
Census tract level SDH data on federal poverty status, education level, median household income, employment status, and health insurance coverage were obtained from the ACS 2013-2017, with 5-year estimates [15]. Those five SDH indicator variables are generally recognized as key determinants that affect women in the United States [16 -18]. The selected SDH variables were categorized by using empirically derived quartiles, determined based on data from all census tracts in the United States and Puerto Rico.

Analysis
This ecological analysis used NHSS data from the 50 states, the District of Columbia, and Puerto Rico to examine distributions of HIV diagnoses by age group, race/ethnicity, transmission category, residence at time of diagnosis, and SDH indicators among women with a diagnosis of HIV infection in 2017. Because a substantial proportion of cases of HIV infection are reported to CDC without an identified risk factor, multiple imputation was used to assign a transmission category [19].
Percentages and rates of HIV diagnoses per 100,000 population were calculated by region for adult women (aged ≥18 years). Distributions were assessed by age group, race/ethnicity, transmission category, residence at the time of diagnosis (urban, suburban, rural, or unknown), and SDH indicator variables in each census tract. The denominators used for calculating age and race/ethnicity-specific rates were computed by applying the 5-year (2013-2017) estimates for age, sex, and race/ethnicity for these areas [20].
The denominator population from the ACS was based on all women aged 18 years and older. ACS uses predetermined age categories and varying criteria for SDH variables, so the denominators differ for some SDH variables. The numerator population was limited to women aged 18 years and older whose HIV infection had been diagnosed and reported with complete residential address information that could be geocoded to the census tract level. Cases or census tracts were excluded if 1) the address was nonresidential (e.g., military base, corrections facility, etc.), 2) no census tract was associated with the case, 3) no SDH information was available for the census tract, or 4) the census tract from the surveillance data could not be matched to a census tract provided by ACS. Regions were based on the U.S. Census Bureau's division of the 50 states [15] and one U.S. territory: Puerto Rico. Area of residence was defined as urban (metropolitan area ≥500,000 population), suburban (metropolitan area of 50,000-499,999 population), or rural (nonmetropolitan population) [14]. SAS version 9.4 (SAS Institute, Inc., Cary, NC) was used to conduct all analyses.

RESULTS
This analysis used data on 6,054 adult women with HIV infection diagnosed in 2017 that included sufficient residential address information for geocoding to census tracts for 5,972 women in the United States and 82 women in Puerto Rico. The overall rate (per 100,000 population) of diagnoses of HIV infection among women in the United States and Puerto Rico was 2.4. Because of the small cell sizes, results stratified by demographic and SDH characteristics for women in Puerto Rico are not discussed.
In terms of the U.S. region, the overall rate of diagnosis was highest in the South (2.4), followed by the Northeast (1.5), the Midwest (1.0) and the West (0.9) ( Table 1). By age, the highest rates were among women aged 35-44 years for the Northeast (3.5), and West (1.9) and among women aged 25-34 years in the Midwest (2.8) and the South (5.7). By race/ethnicity, Black women had the highest HIV diagnosis rates in every region: Northeast (22.5),Midwest (17.7), South (24.4), and West (18.2). By transmission category, in every region, the largest percentages of HIV diagnoses were among women with HIV infection attributed to heterosexual contact: Northeast (83.3%), Midwest (83.3%), South (90.1%), and West (80.9%). By residence area, in every region, the largest percentages were in urban areas: Northeast (92.3%), Midwest (79.5%), South (76.2%), and West (88.3%) ( Table 1).  Race

Examination of Geographic Differences in Social Determinants
The   Abbreviation: Asterisk (*) indicates cell too small (<12 diagnoses during the year) and data has been suppressed. Dash (-) indicates absence of denominator data and rate is not calculated. Note: Rates are per 100,000 population. Rates are not calculated by transmission category because of the lack of denominator data. The American Community Survey does not provide age-and sex-specific social determinants of health variables stratified by race and ethnicity for racial groups except Whites (i.e., White, not-Hispanic/Latino). Therefore, the denominators used for computing race-specific rates may include Hispanics/Latinos and should be interpreted with caution. By race/ethnicity and age, among white women, those aged 25-34 years had the highest rates of HIV diagnoses (in every region) ( Table 2). Among Black women, those aged 25-34 years in the Midwest and South and those aged 35-44 years in the Northeast and West had the highest rates. Among Hispanic/Latino women, those aged 45-54 years had the highest rates in all regions except for the South, where those aged 25-34 years had the highest rate ( Table 2). By race/ethnicity and transmission category, the largest percentages of HIV diagnoses were among women with HIV infection attributed to heterosexual contact, regardless of region or race/ethnicity. By race/ethnicity and residence area, the largest percentages were among women who lived in urban areas.

Examination of Geographic Differences in Social Determinants
The  By SDH for each race/ethnicity group and across all regions, the highest rates of HIV diagnoses were among women who lived in census tracts where at least 19% of residents had incomes below the federal poverty level. The one exception was Hispanic/Latino women in the Midwest, for whom the highest rate was among those women who lived in census tracts where at least 11% but fewer than 19% of the residents had incomes below the federal poverty level. Women with the highest HIV rates lived in census tracts where at least 18% of residents had less than a high school diploma. An exception was Black women in the West, for whom the highest rates were among women living in census tracts where at least 11% but fewer than 18% of the residents had less than a high school diploma. Another exception was Hispanic/Latino women in the Midwest, for whom the highest rates were among women living in census tracts where at least 6% but fewer than 11% of the residents had less than a high school diploma. Women with the highest HIV diagnosis rates also lived in census tracts where the median household income was less than $40,000, and at least 16% were without health insurance. The exception was Hispanic/Latino women in the Northeast and Midwest, among whom the highest diagnosis rates were in census tracts where at least 10% but fewer than 16% of the residents were uninsured. Among White women, those who lived in census tracts where at least 6% of the residents were unemployed accounted for the highest HIV diagnosis rates in every region; no clear pattern was observed for Black and Hispanic/Latino women ( Table 2).

Below the federal poverty level (%)
By residence area and race/ethnicity, the largest number of HIV diagnoses (2,984) was among Black women in urban areas ( Table 3). By age and residence area, the largest percentages of HIV diagnoses were among women aged 25-34 years, regardless of race/ethnicity (except for Hispanic/Latinos in rural areas). By residence area and transmission category, the largest percentage was among women with HIV infection attributed to heterosexual contact, regardless of race/ethnicity. By residence area and SDH, regardless of race/ethnicity, the highest percentages were among those who lived in census tracts where at least 19% of the residents lived below the federal poverty level (except White women in rural areas), at least 18% of the residents had less than high school diploma (except White women in suburban areas), the median household income was less than $40,000 a year (except White women in urban areas) and 16% or more of the residents did not have health insurance or a health coverage plan (Table 3). Table 3. Count and percentages of diagnoses of HIV infection among women, by race/ethnicity, area of residence and selected characteristics, 2017─United States, the District of Columbia and Puerto Rico.

DISCUSSION
The results of this study are supported by findings in the previous research [ 21 -24]. Overall, women who lived in areas with the highest poverty, highest unemployment, lowest incomes, and those without health insurance coverage were more likely to have higher rates of HIV infection than did women who lived in the higher favorable quartiles. The South followed the national trends in having the highest number of HIV diagnoses and the highest rates of HIV infection among women in the SDH examined. Overall, women in the least favorable quartiles in every region, regardless of race/ethnicity, tended to have the highest diagnosis rates. However, patterns in the SDH categories varied among Hispanic/Latino women in the Midwest. Across regions, higher rates of diagnoses were seen among White women aged 25-34 years. Among Blacks and Hispanics/Latinos, higher rates were seen in women aged 35 years and older for certain regions. Additionally, Black women in every area of residence had the highest percentages of HIV diagnoses in the least favorable SDH quartiles. Overall, across all races/ethnicities and regions, rates of HIV diagnoses were highest among women residing in the census tracts with the lowest median household income and lacking health insurance coverage.
The results of this study showed that women who receive an HIV diagnosis, regardless of race/ethnicity and region, are generally in the least favorable quartiles of each SDH analyzed (barring a few exceptions). Research has shown an association between poor health and living in areas with high poverty levels, having low income, lacking health insurance coverage, lacking education, and lacking employment [25 -27]. This is more pronounced among women, who often put their health needs aside to meet other priorities to survive and take care of their families [28,29]. When race is considered, we find that there are more pervasive disparities in access to health insurance coverage among minority women [30,31]. The Coverage and Access to Care Under the Affordable Care Act reports that overall, 86% of Black women (ages 18-64 years) had health insurance in 2017, compared with nearly 92% of white women [32]. Moreover, Black women are more likely to hold low-wage jobs that might not provide adequate health benefits [33].
The results of this study also showed that the highest rates of HIV diagnoses among white women were seen in a younger age group (25-34 years). In contrast, Black and Hispanic/Latino women mostly had the highest rates of HIV diagnoses in older age groups. This might be a result of Black and Hispanic/Latinos receiving a late HIV diagnosis. Previous studies show that Black and Hispanic/Latino women tend to present at later stages in their HIV diagnosis [34 -36]. This could be attributed to varying attitudes towards sexual and reproductive health among younger women of certain racial/ethnic origins. Among Black and Hispanic/Latino women, stigma might restrain young unmarried women from going for an HIV test [37,38]. The fear of being HIV positive and fear of the social consequences, such as rejection by loved ones, loss of employment or housing, and the fear of discrimination and violence, can lead to delays in testing for HIV [37 -39]. This concurs with a study involving Black and Latino women at risk for HIV, which showed that stigma and the perception that these women were engaging in risky behaviors influenced their decision to visit a clinic to receive an HIV vaccine in a vaccine trial study [39]. Through the decades, Black and Hispanic/Latino women also have had less access to treatment and prevention services [40 -42].
The results of this study showed that Black women in every type of residence area had the highest HIV diagnosis percentages in the least favorable quartile of each SDH. However, this does not hold the same for White and Hispanic/Latino women. This might be because Black women often experience multiple SDH that might negatively affect their health. For instance, Black women experience poverty at higher rates than do Black men and women from all other racial/ethnic groups [42,43]. One study reported that non Hispanic Black women were most likely to be single heads of households with family members present (27.5%), and non Hispanic Asian and non-Hispanic White women were least likely (7.5% and 9.4%, respectively) [43]. More than 30% of female single heads of households with family members had household incomes below the poverty level, contributing further to worse health outcomes [43]. However, White women had the highest percentages of HIV diagnoses in the second most favorable unemployment level, regardless of region. This suggests that being unemployed for White women is not necessarily a factor for poor health. It is possible that White women are more likely to be employed than women of other races [2,26], but we cannot explain why they had higher rates of HIV diagnosis in the more favorable unemployment census tracts.
The results of this study showed that regardless of race/ethnicity, women who lived in a census tract with the lowest median household income and the highest percentage of uninsured residents (except for Hispanic/Latino women in the Northeast and Midwest) generally had the highest HIV diagnosis rates. Several studies have shown that women with low incomes might have poorer health outcomes and are unable to afford health insurance coverage might have poorer health outcomes [44 -46]. Previous research has given important insights into how health care coverage affects health care use, disease treatment and outcomes, self-reported health, and mortality [47]. Women in worse health and those with low incomes, or those uninsured, are more likely than others to delay or forgo health services because of the costs associated with health coverage. Women in low-income households tend to decline needed health care screenings to keep income within the household [27,28]. There is also strong evidence that health insurance coverage increases access to preventive services, such as pre-exposure prophylaxis, which can directly maintain or improve health and reduce the chances of being infected with HIV [47,48].
This analysis is subject to at least four limitations. First, diagnoses of HIV do not represent incidence of HIV or new infections. The time from infection to diagnosis will differ with each woman and the residence at HIV diagnosis might not be the same as when HIV was acquired. Second, data were limited to women with a complete residential address that could be geocoded; therefore, the results might not be representative of all women receiving an HIV diagnosis in the United States in 2017. Third, because SDH information is not available at the individual level, census tract-level data were used as a snapshot of the environment in which women lived at the time of HIV diagnosis. Conclusions and findings should be interpreted with caution and not inferred at the individual level. Fourth, some SDH indicators (e.g., lack of health coverage) are associated with low income and poverty in the United States [27,49,50], but correlations between SDH indicators were not analyzed for this report.

CONCLUSION
This publication is the first of its kind to examine the distribution of census tract-level SDH among women receiving an HIV diagnosis in the 50 states, DC, and Puerto Rico. The results show that the burden of HIV among women by region is affected by race/ethnicity and SDH. By identifying the SDH that are strong indicators for a diagnosis of HIV among select populations of women, and the distribution of HIV by geography, targeted interventions can be tailored to those specific populations, which might yield better health outcomes. Future interventions should take into consideration SDH, such as income and health insurance coverage, among women with respect to race/ethnicity, region and area of residence. Examining these will allow appropriate allocation of resources to the geographic areas where they are needed the most, thereby meeting the objectives outlined in the federal initiatives.

AUTHORS' CONTRIBUTIONS
All authors contributed to the design of the study, analysis, interpretation, and write up of the manuscript. All authors read and approved the final manuscript.