Population Density and Health Outcomes in Women with HIV in the Southern United States: A Retrospective Longitudinal Analysis

Abstract

Purpose:

Published studies have revealed challenges for people with human immunodeficiency virus (HIV) living in rural areas compared to those in urban areas, such as poor access to HIV care, insufficient transportation, and isolation. The purpose of this study was to examine associations between population density and multiple psychosocial and clinical outcomes in the largest cohort of women with HIV (WWH) in the United States.

Methods:

Women’s Interagency HIV Study (WIHS) participants from Southern sites (n = 561) in 2013–2018 were categorized and compared by population density quartiles. The most urban quartile was compared with the most rural quartile in several psychosocial and clinical variables, including HIV viral load suppression, HIV medication adherence, HIV care attendance, depression, internalized HIV stigma, and perceived discrimination in healthcare settings.

Results:

Although women in the lowest density quartile were unexpectedly more highly resourced, women in that quartile had greater odds of not attending an HIV care visit in the last six months (odds ratio [OR] = 0.64, 95% confidence interval [CI] [0.43–0.95]), yet higher odds for having fully suppressed HIV when compared to women in the highest density quartile (OR = 1.64, 95% CI [1.13–2.38]). Highly urban WWH had greater likelihood of unsuppressed HIV, even after controlling for income, employment, and health insurance, despite reporting greater HIV care adherence and similar medication adherence.

Discussion:

Further investigation into the reasons for these disparities by population density is needed, and particular clinical attention should be focused on individuals from high population density areas to help maximize their health outcomes.

Introduction

Although the incidence of human immunodeficiency virus (HIV) syndrome was initially higher in the highly urban areas of the Northern and Western U.S., over the past 20 years there have been higher rates of HIV infection in the more rural southeastern states.^1
–3 According to Centers for Disease Control and Prevention data, the states with the highest rates of new HIV diagnoses are in the Southeastern region of the United States, which is also where 42% of deaths among people with HIV were reported in 2020.⁴ Specifically among women, in 2021, the states with the highest HIV mortality rates (outside of New York and Maryland) were in the South, as were the states with the highest rates of new HIV diagnoses.^5,6

Multiple studies have concluded that people with HIV in rural settings have more barriers to antiretroviral treatment (ART) than those in non-rural settings. These include personal barriers such as greater social isolation from others living with HIV,⁷ as well as greater risk for depression and more problems coping with their HIV diagnoses.^8

–11 In addition, there are care-related barriers, such as greater distances to HIV care¹² decreased utilization of health care resources,¹³ reduced availability of health care services experienced in treating HIV,^14,15 lack of transportation,^16
–18 and decreased care retention.^18

–21

Concerns specific to women with HIV (WWH) include internalized stigma and a perception of discrimination in health care settings,^9,22 hesitance in HIV disclosure,²³ and a lower rate of viral suppression, especially in women of color.²⁴ Deficits related to women’s health include poorer cervical cancer screening,²⁵ and higher sexually transmitted infection (STI) prevalence.²⁵ In contrast, some studies showed more challenges for women in urban areas such as lower rates of viral load suppression,²² lower income, education, and health insurance,^22,26 and transportation barriers.²⁶

Previous research examining the association between population density and HIV stigma have yielded contradictory evidence. While one study of men and women in Georgia found that living in less densely populated communities was correlated with higher internalized HIV stigma,⁹ another study concluded that community size was not associated with internalized HIV stigma,²³ although one aspect of HIV stigma, “disclosure concerns,” was reported to be higher in rural women compared with their metropolitan counterparts. In a cross-sectional analysis of the baseline visits for Southern site participants of the Women’s Interagency HIV Study (WIHS), Edmonds and colleagues also found that there was greater internalized HIV stigma and perceived discrimination in health care settings reported by women living in non-urban settings.²²

To improve our understanding of how urbanicity can impact health outcomes in WWH so that barriers can be identified and interventions developed, we analyzed longitudinal data from the WIHS. This study used data collected from WWH attending sites in five Southern states over a five-year period. In the aforementioned cross-sectional analysis at the baseline visit,²² the authors concluded that clinical outcomes such as CD4 count and HIV viral load suppression were not associated with population density, though women who lived in the lower density quartiles were more likely to report perceived discrimination and internalized HIV stigma than women in higher density quartiles. This longitudinal exploration examines similar questions using data over a 5-year period to explore whether the associations between the population density of one’s area of residence and health outcomes changed over time.

Methods

Sample. The WIHS, which began in 1993, was a large longitudinal prospective cohort study to explore the impact of HIV on women in the United States.^27
–29 In 2019, it combined with the similar and even more long-standing men’s study, the Multicenter AIDS Cohort Study (MACS), to become the MACS/WIHS Combined Cohort Study (MWCCS).³⁰ Although the original WIHS sites were in major cities in Northeastern or Western U.S., four Southern sites in Florida, North Carolina, Georgia, and Alabama/Mississippi (joint site) were added in 2013 to reflect the higher HIV incidence and prevalence in the Southeastern U.S.²⁷ In 2013, WWH or comparable women without HIV between the ages of 25 and 60 years were eligible to enroll in the WIHS in these Southern sites. Women without HIV were required to have risk factors for HIV, meaning that they needed to report at least one substance use or sexual behavior indicating higher HIV risk, or to have had a sexual partner living with HIV or having those same higher risk behaviors, in the past 5 years. Participants attended semiannual study visits in order to provide interview, anthropometric, imaging, and laboratory data as well as biological specimens. Ethics approval for data collection was given by the Institutional Review Boards at each local site. This particular analysis was exempt from informed consent, given that it was secondary analysis of de-identified data.

This analysis included participants only from these Southern sites, as they had more variability in population density than participants from the original sites. Starting in 2013, the WIHS collected home addresses of all participants annually. WWH at Southern sites who consented to geocoding and had at least one geocoded address available at any point during the follow-up period (2013–2019) were included in this analysis.

For this sample, out of 566 unique women at Southern sites, 561 had at least one geocoded address available at any point during the follow-up period. The demographics of the sample of 561 WWH at the baseline visit, as well as a summary of demographics at all visits, are summarized in Table 1. At baseline, the median age was 44 years, (interquartile range [IQR]: 36, 51) and 83% were African American. Most had high school education or less (64%), and annual household incomes ≤ $12,000 (55%), which was approximately the federal poverty line. For all visits over the 5-year period, the mean age was 47 years (IQR: 39, 53), and education and income did not change. The majority reported having some form of health insurance (85% at baseline; 93% for all visits).

Table 1.

Characteristics of the Sample at First Visit (n = 561 and All Visits (n = 3,265); Population Density Quartiles at First Visit

		Median (Q1, Q3) or N (% of total)
Characteristic	Overall first visit	Overall all visits	Q1^a (n = 142)	Q2(n = 139)	Q3(n = 139)	Q4(n = 141)	p-value
Age (years)	44 (36, 51)	47 (39, 53)	43 (35, 51)	43 (35, 50)	46 (38, 52)	46 (38, 52)	0.23
Race/ethnicity
African-american (non-hispanic)	466 (83.6%)	2,720 (83.6%)	111 (78.2%)	125 (89.9%)	122 (87.8%)	108 (76.6%)	<0.0001^***
White (non-Hispanic)	53 (9.4%)	304 (9.3%)	21 (14.8%)	9 (6.5%)	13 (9.4%)	10 (7.1%)
Other	42 (7.5%)	228 (7.0%)	10 (7.0%)	5 (3.6%)	4 (2.9%)	23 (16.3%)
Education
Some college or more	204 (36.4%)	1,216 (37.4%)	61 (43.0%)	52 (37.4%)	52 (37.4%)	39 (27.7%)	0.12
Completed high school	183 (32.6%)	1,032 (31.7%)	47 (33.1%)	43 (30.9%)	46 (33.1%)	47 (33.3%)
Less than high school	174 (31.0%)	1,004 (30.9%)	34 (23.9%)	44 (31.7%)	41 (29.5%)	55 (39.0%)
Annual household income
<$6,000	119 (21.3%)	618 (19.7%)	30 (21.3%)	24 (17.5%)	21 (16.3%)	39 (28.5%)	0.029^*
$6,001–$12,000	188 (33.6%)	1,099 (35.1%)	38 (27.0%)	43 (31.4%)	48 (37.2%)	54 (39.4%)
$12,001–$24,000	155 (27.7%)	866 (27.7%)	37 (26.2%)	44 (32.1%)	40 (31.0%)	30 (21.9%)
$24,001–$36,000	58 (10.4%)	346 (11.1%)	22 (15.6%)	15 (10.9%)	12 (9.3%)	9 (6.6%)
>$36,001	40 (7.1%)	201 (6.4%)	14 (9.9%)	11 (8.0%)	8 (6.2%)	5 (3.6%)
Insurance
Yes	475 (84.7%)	3032 (93.4%)	116 (81.7%)	118 (84.9%)	116 (83.5%)	123 (88.5%)	0.44
ART adherence > 95%
Yes	458 (84.7%)	2,624 (86.5%)	110 (85.9%)	96 (82.8%)	103 (86.6%)	92 (82.9%)	0.78
STI reported or diagnosed at visit
No	487 (86.8%)	3,050 (93.9%)	129 (90.8%)	122 (87.8%)	114 (82.0%)	122 (87.1%)	0.17
Self-reported abnormal gynecological exam
No	470 (83.8%)	2,791 (85.8%)	117 (82.4%)	123 (88.5%)	112 (80.6%)	118 (83.7%)	0.32
No depressive symptoms
CES-D score: 0–15	309 (55.3%)	2,048 (63.3%)	80 (56.3%)	82 (59.0%)	78 (56.9%)	69 (50.0%)	0.47
History of emotional abuse
No	525 (94.3%)	3,132 (96.7%)	134 (94.4%)	130 (93.5%)	129 (94.2%)	132 (95.0%)	0.96
Social support^e	4.1 (3.1, 4.9)	4.1 (3.1, 5.0)	4.1 (3.3, 5.0)	4.0 (3.0, 4.9)	4.0 (3.0, 4.6)	4.2 (3.0, 5.0)	0.21
Social support category^f
Adequate to high (> 3)	418 (78.1%)	2,178 (80.1%)	105 (77.8%)	108 (77.1%)	95 (79.2%)	110 (78.6%)	0.98
Ever sexual abuse
No	547 (97.5%)	3,190 (98.6%)	141 (99.3%)	135 (97.8%)	132 (96.4%)	134 (96.4%)	0.34
Utilization of HIV care^b
Yes	475 (91.0%)	2,385 (88.4%)	123 (93.2%)	116 (85.3%)	113 (94.2%)	123 (91.8%)	0.05
Internalized HIV stigma^c	2.1 (1.6, 2.6)	2.0 (1.4, 2.4)	2.3 (1.6, 2.9)	2.1 (1.6, 2.6)	2.1 (1.4, 2.4)	2.1 (1.7, 2.6)	0.25
Internalized HIV stigma category^d
>1.5	429 (76.7%)	1,680 (68.0%)	104 (75.9%)	100 (75.8%)	97 (73.5%)	106 (79.7%)	0.69
Perceived discrimination in health care settings^g
No	442 (82.9%)	2,130 (86.2%)	103 (75.2%)	108 (81.8%)	113 (86.3%)	118 (88.7%)	0.018^*
HIV viral load
< 200 copies/mL	397 (75%)	2,624 (83.8%)	107 (84.3%)	99 (73.3%)	95 (72.0%)	96 (71.1%)	0.05
CD4 count
> 200	497 (89.7%)	3,019 (94.0%)	130 (93.5%)	120 (87.6%)	126 (91.3%)	121 (86.4%)	0.18

Percentages are calculated using records with non-missing data.

Q1 = lowest population density quartile – Q4 = highest population density quartile.

Data from Visit 2 were used, given the large amount of missing data in Visit 1.

Mean score of an adapted 7-item “negative self-image” subscale of the adapted HIV Stigma Scale, higher score indicating higher internalized HIV stigma.

Mean score ≥ 1.5 (median) “negative self-image” subscale of adapted HIV Stigma Scale, indicating higher internalized HIV stigma.

Data from Visit 2 were used, given large amount of missing data in Visit 1; mean emotional and tangible social support on the Medical Outcomes Study Social Support Scale (MOS-HIV).

Mean emotional and tangible social support ≥ 3 on the MOS-HIV.

Perceived discrimination in health care settings using a question drawn from the aforementioned HIV Stigma scale.

p < 0.05, **p < 0.01, ***p < 0.001.

Categorical variables were compared with chi-square tests and continuous variables were compared with Kruskal–Wallis tests.

ART, antiretroviral therapy; STI, sexually transmitted infection; CES-D, Center for Epidemiological Studies Depression Scale.

Measures

Classification of participant urbanicity

ArcGIS, a geospatial software to view, edit, manage, and analyze geographic data, was used to geocode address data. Each participant was assigned a Federal Information Processing Standard Code (FIPS) code after the latitude and longitude of their addresses were matched with census tracts. The FIPS codes were linked to population density data from various versions of the 5-year American Community Survey (ACS),³¹ an ongoing general household survey, to obtain the census tract population density (persons per square mile) for each person-visit in the analysis. Visits in 2013–2014 were linked to the 2012–2016 ACS, 2015 visits to 2013–2017, 2016 visits to 2014–2018, and 2017–2019 visits to 2015–2019. The quartile cut points were based on the first geocoded visit for each woman in the dataset, yet the assignment of quartile category was time-varying across visits. We used the first geocoded visit for each woman in the dataset to generate the following cut points for population density quartiles: minimum, 10; quartile 1 (Q1), 1,400; median, 2,700; quartile 3 (Q3), 4,600; maximum, 33,000 (all values were rounded to the nearest 100, except the minimum, which was rounded to the nearest 10, and the maximum, which was rounded to the nearest 1,000). We then classified the population density at each person-visit into these four categories.

Demographic variables

Demographic variables were (a) age, (b) education (<high school, completed high school, or some college or more), (c) annual household income (≤$6,000, $6,001–$12,000, $12,001–$24,000, $24,001–$36,000, >$36,000), (d) race/ethnicity (White non-Hispanic, African-American non-Hispanic, Other), and (e) health insurance (yes/no).

Psychosocial variables

Psychosocial variables were (a) depressive symptoms, (b) history of domestic violence, (c) history of sexual abuse, (d) social support, (f) HIV stigma, (g) discrimination, and (h) years since the self-reported date of the first HIV-positive test.

Depressive symptoms: We utilized the Center for Epidemiological Students Depression (CES-D) Scale, which comprised 30 yes or no questions. If a participant scored 16 or more, this indicated clinical risk for depression. The higher the score, the more depressive symptoms are reported.

History of domestic abuse: Participants reported a history of domestic abuse by responding yes or no to one question: yes if answering affirmatively to “Have you ever experienced serious physical harm by another person? By that I mean were you ever hurt by a person using an object or were you ever slapped, hit, punched, or kicked?”

History of sexual abuse: History of sexual abuse was indicated by responding yes/no to one question, yes if answering affirmatively to “At any time in your life, has anyone ever pressured or forced you to have sexual contact? By sexual contact I mean them touching your sexual parts, you touching their sexual parts, or sexual intercourse.”

Social support: Social support was derived from the mean social support score (representing emotional and tangible social support) from the Medical Outcomes Study Social Support Scale³² (range: 1–5). This was also used as a dichotomous variable, where adequate to high social support was indicated by values >3.

HIV stigma: Participants endorsed Strongly Agree to Strongly Disagree in a 5-point Likert scale in seven items reflecting the “negative self-image” subscale of the HIV Stigma Scale. Sample statements included “Having HIV/AIDS makes me feel that I’m a bad person” and “I feel guilty because I have HIV/AIDS.” The score represents the mean of the seven questions, each measured on a scale of 1–4, with higher scores representing higher internalized HIV stigma.³³ This was also used as a dichotomous variable, where high internalized HIV stigma was indicated by scores at or above the median of 1.5.

Discrimination: Perceived discrimination in health care settings was derived using one question drawn from the validated discrimination scale, measured on a scale of 1–4, with higher scores representing higher perceived discrimination:³⁴ “I feel discriminated against in health care settings because of my HIV status.” This was also used as a dichotomous variable, where no perceived discrimination in health care settings was indicated by a neutral response or disagreement (i.e., disagree or strongly disagree) to the question.

Years since self-reported date of first HIV-positive test: This was derived based on when the participant reported their first HIV-positive test result.

Clinical outcome variables

Clinical outcomes were: (a) HIV viral suppression (yes/no, yes if < 200 copies/mL), (b) absolute CD4 count, absolute CD4 category (< 200; ≥ 200), (c) ART adherence in the past 6 months if receiving ART at the visit (yes/no; yes if medications taken ≥ 95% of the time), (d) utilization of HIV care: received HIV care in the past 6 months (yes/no), (e) self-reported abnormal cervical screening result or treatment for a gynecological condition in the past 6 months (yes/no), (f) self-reported gonorrhea, syphilis, chlamydia, pelvic inflammatory disease, trichomonas, genital herpes, or warts in the past 6 months, or clinical diagnosis of any of these STIs during the study visit (yes/no). All variables were time-varying except race/ethnicity, education, lifetime history of sexual abuse, and lifetime history of domestic violence.

Data Analysis

For modeling, we numbered each participant’s attended visits during the follow-up period as 1–8 based on when the participant had their first WIHS visit, and we accounted for missing data by using multiple imputation by fully conditional specification, creating 10 versions of the dataset. Each imputed dataset was analyzed via logistic regression with generalized estimating equations (GEEs) and robust standard errors, and we assumed an exchangeable correlation structure. The imputed datasets were appropriately statistically combined to yield adjusted odds ratios (ORs). Categorical variables were compared with chi-square tests and continuous variables were compared with Kruskal–Wallis tests. All analyses were performed in SAS 9.4 (SAS Institute Inc., Cary, NC).

We used logistic regression models to compare outcomes by population density quartiles utilizing data from Q1 and quartile 4 (Q4), while controlling for other variables theoretically related to each outcome, using the highest density quartile (most urban) as the referent group in all analyses. Table 2 includes the control variables used for each health outcome of interest and are described within the Results section for each analysis. GEE was needed because we were interested in population average effects as well as to account for the repeated measures nature of the data.

Table 2.

Outcomes by Population Density Quartile, with High Density Quartile as Referent Quartile

Characteristic^a	Control variables	N visits	Odds ratio	95% confidence interval
Viral load suppression^b	Insurance, income, race, adherence, internalized HIV stigma	3,180	1.64	1.35–2.38^**
CD4 count > 200	Insurance, race, adherence, internalized HIV stigma, depression, social support	3,186	1.36	0.78–2.35
ART adherence > 95%	Insurance, income, race, internalized HIV stigma, depression, social support, utilization of care	3,180	1.00	0.66–1.50
Utilization of HIV care	Insurance, income, race, internalized HIV stigma, depression, social support	3,242	0.64	0.43–0.95^*
STI reported or diagnosed at visit	Income, race, sexual abuse, emotional/domestic violence, depression	3,259	0.54	0.33–0.88^*
Abnormal gynecological exam in the last 6 months	Insurance, race, STI, CD4, viral load, utilization of care	3,250	0.97	0.67–1.40
Internalized HIV stigma^c	Income, race, depression, social support, utilization of care, years since first positive HIV test	3,242	1.05	0.76–1.45
Perceived discrimination^d	Education, race, depression, internalized HIV stigma, social support, utilization of care	3,248	1.20	0.78–1.83
Social support^e	Education, race, depression, internalized HIV stigma	3,252	0.96	0.66–1.40
Depressive symptoms^f	Education, race, social support, internalized HIV stigma, utilization of care	3,248	0.77	0.58–1.03

Odds of outcome in the lowest population density quartile as compared to the highest population density quartile.

Viral suppression = HIV viral load < 200.

mean score ≥ 1.5 (median) “negative self-image” subscale of adapted HIV Stigma Scale, indicating higher internalized HIV stigma.

perceived discrimination in health care settings using a question drawn from the aforementioned HIV stigma scale.

mean emotional and tangible social support < 3 on the Medical Outcomes Study Social Support Scale, indicating lower support.

depressive symptoms using the Center for Epidemiological Studies Depression Scale (score ≥ 16).

p < 0.05.

p < 0.01.

ART, antiretroviral therapy; STI, sexually transmitted infection.

Results

There was a total of 3,265 visits with a geocoded residential address, with a median of 6 visits per participant (IQR: 5, 7). There were 898 visits in the lowest population density quartile, 765 in the next highest quartile, 870 in the second highest quartile, and 732 in the highest (most urban) quartile. The maximum population density of the highest density quartile was 23 times higher than the maximum of the lowest density quartile.

Race/ethnicity, annual household income, and internalized HIV stigma differed by population density quartile (Table 1), with a higher percentage of non-Hispanic African Americans in Q2 and Q3 than in the most rural quartile (Q1) or the most urban quartile (Q4). The quartile with the highest number of non-Hispanic Whites was Q1 (14.8%), whereas Q4, the most urban quartile, had the highest percentage of Others, who were primarily Hispanic participants (16.3%). The quartile with the highest proportion of the lowest-income participants was the most urban quartile (Q4; 68% reporting ≤ $12,000/year), and the highest income category (≥ $36,001 per year) was most often reported in the two lowest population density quartiles (10% in Q1 and 8% in Q2 as compared with 6% in Q3 and 4% in Q4).

At the first visit, 85% of participants reported ART adherence of ≥ 95%, 91% reported having attended an HIV care appointment in the previous 6 months, and 75% had suppressed HIV (< 200). Considering all visits over the 5-year period, 86% reported ≥ 95% adherence, 88% reported attendance at an HIV care appointment in the previous 6 months, with 84% exhibiting suppressed HIV. While adherence and viral suppression increased at least slightly when including all visits, report of utilization of HIV care in the past 6 months declined. There was no significant difference between quartiles, though there was a trend (p = 0.05) for viral suppression to be greater in the most rural quartiles (84.3% at Q1) as compared to the more urban quartiles (71.1% at Q4).

A large minority (44%) of the population expressed depressive symptoms (CES-D score ≥ 16) at their first visit, and 37% of all visits displayed this same level of depressive symptoms. In addition, at first visit, 76% of the population had internalized stigma scores higher than the median, and 17% endorsed perceived discrimination in the healthcare setting, whereas when grouping all visits, the number reporting higher internalized stigma dropped to 68%, and 14% endorsed perceived discrimination. Thirty-two percent reported inadequate social support at the first visit, although this reduced to 20% when grouping all visits. Thus, over the time of data collection, depression, internalized stigma and perceived discrimination decreased, and social support increased.

Other demographic, psychosocial, and HIV outcomes were similar between quartiles. Thirteen percent reported a previous STI at the first visit, with an STI reported at only 6% of all visits. Sixteen percent reported an abnormal gynecological exam at the first visit, and 14% of all visits. History of emotional or sexual abuse was low at 6% and 2.5%, respectively, at the first visit, and 3% and 1.4% of all visits.

The adjusted OR for HIV viral suppression was 1.64 (95% confidence interval [CI] [1.13–2.38]) for the lowest density quartile compared to the highest density quartile, controlling for health insurance, annual household income, race/ethnicity, ART adherence, and internalized HIV stigma. We also found that WWH were less likely to report HIV care utilization at a given visit if they were in the lowest population density quartile than if they were in the highest (most urban) population density quartile (OR = 0.64, 95% CI [0.43–0.95]), and that was also the case for reporting or having a clinically diagnosed STI (OR = 0.54, 95% CI [0.33–0.88]).

In contrast to a previous analysis,²² no significant difference related to population density was found in the odds for internalized HIV stigma (controlling for income, race, depression, social support, utilization of care, and years since first positive HIV test) and perceived discrimination (while controlling for education, race, depression, internalized HIV stigma, social support and utilization of care) when comparing the lowest density quartile to the highest density quartile, and when considering all visits.

Discussion

This study aimed to advance previous findings on differences in clinical and psychosocial outcomes between urban and non-urban women in WIHS Southern sites,²² by adding data from 2013–2019. Consistent with the previous study, we found less frequent utilization of health care for WWH in the most rural quartile as compared to those in the most urban quartile, yet we found that HIV viral suppression was more likely for women in the most rural as compared to the most urban quartile. When examining predictors for utilization of care, those in the most rural quartile had lower odds of HIV care utilization while adjusting for theoretically driven confounders (insurance, income, race, internalized HIV stigma, depression, social support). The odds of ART adherence were not significantly affected by population density, when controlling for insurance, income, race, internalized HIV stigma, depression, social support, and utilization of care. This disconnect between self-reported ART adherence being unaffected by population density, yet HIV viral suppression being greater and utilization of care being less in the most rural quartiles is somewhat counterintuitive. If considering the impact of structural factors on health, it could be speculated that the higher income in the most rural quartile translated into better access to healthcare providers outside of clinic visits, through a sense of agency in using electronic messaging apps or telephone consultation when distance from the clinical site made travel more difficult, though income was controlled for in these analyses. Further research is needed to understand these relationships.

We found that comparing the lowest to the highest population density quartiles was not associated with either internalized HIV stigma and perceived discrimination in health care settings, despite it being associated with these factors at the baseline visit as reported in the previously published report.²² One interpretation of this finding is that these participants became more comfortable in their self-perceptions about living with HIV as a result of living longer with HIV, as well as a result of being involved in a study that required semiannual contact with staff who were openly welcoming to those living with HIV, with visits centered around them talking about HIV and the issues it brought to their lives. Study visits took place in health care settings that were either at an academic medical setting or associated with one. Over time, it may be that the participants began to feel more comfortable in health care settings as a result of their involvement in the WIHS, and thus perceived less discrimination. Another explanation could be that those with higher internalized HIV stigma or perceived discrimination in health care settings dropped out, thus creating a lower percentage of the total. While this could be the case, retention in the WIHS was high, and thus, this explanation is less likely.

The lower odds of reported or clinically diagnosed STIs for women in the least urban areas (versus most urban areas) may have been due to a higher prevalence of STIs in the more urban areas, or a greater availability of STI testing in those areas, and thus a higher rate of reported STI diagnoses at study visits.

No differences related to population density were found for CD4 count > 200, abnormal gynecological findings, social support or depressive symptoms, when examining these outcomes by population density quartiles and adjusting for theoretically driven confounders. While some reports of those living with HIV in rural environments have reported otherwise for depression and social support,^3,35 the population of individuals in the lowest population density quartile in this study may have had different characteristics than those studies that were focusing on very rural individuals living with HIV. Although the maximum population density in the most rural quartile in this study was 23 times lower than the maximum population density in the most urban quartile, the representation of truly rural individuals was low.

A limitation of this study is that most women in the lowest population density quartile were not truly rural, but rather resided in what could be referred to as small cities, although by the 2010 U.S. Census Bureau definition of urban areas as having greater than 2,500 individuals,³⁶ the population densities per square mile in the census tract do not indicate an urban area. This precludes the generalization of these results to the truly rural experience. In addition, although the demographics of women in the WIHS in the southern sites reflect the demographics of WWH in the Southeastern United States, their experience in the WIHS may make our results less generalizable to all WWH in the five states where women in this analysis attended study visits. In addition, because only the southern WIHS sites were used for this study in order to have sufficient variability in population density, our results may only be generalizable to the Southern United States. Results may have differed if the low-density category included a higher number of truly rural individuals, and if variables such as isolation could have been measured.

At the same time, being able to report on a substantial amount of longitudinal data in a large number of Southern WWH in relation to their population density is an important contribution to the literature, and provides information from an extended time period. This information, especially comparing the baseline visit to all visits over a 5-year period, can give a greater understanding of the effects of population density as an environmental characteristic. In addition, the use of ArcGIS allows for continuous measurement of population density, allowing for a fine-tuned breakdown of population density rather than a simple categorization of rural and urban.

Conclusions

In a population of southeastern WIHS participants, this study demonstrates outcomes related to living in the least population-dense quartile as compared with living in the highest density quartile, where the maximum population density was 23 times lower than the maximum density of the most urban area. Living in the lowest population density quartile (compared with living in the highest density quartile) was associated with a higher likelihood of HIV viral suppression and a lower likelihood of reporting or being diagnosed with an STI, although utilization of HIV care was also lower for WWH in this quartile, even after accounting for annual household income, health insurance, and internalized HIV stigma. Because participants in the least dense areas reported higher annual household incomes than those in the highest density areas, we speculate that these more highly resourced women may have had a greater access to providers outside of HIV care visits, such as through e-mails or electronic health record messaging systems. This may have led to the higher proportion of viral suppression among WWH (84% versus 71% for women in the most urban areas).

In addition, it appears that living longer with HIV, or longer exposure to clinical/research sites that are accepting and allowing open communication about the challenges of living with HIV, may mitigate the sense of internalized HIV stigma, perceived discrimination, social support, and depressive symptoms over time. While this is valuable information for those who have lived long with HIV, it may indicate the need for clinical and research sites to focus support on individuals newly diagnosed to assure that they are able to communicate openly with supportive staff. While we do not include data about the length of time living in each of the population density quartiles, which would have allowed us to explore whether the length of exposure to the specific environment confers greater adversity or benefit to these outcomes, that would be an interesting next step. In addition, further exploration into these data could determine if the decrease in internalized HIV stigma over time could have resulted from individuals with higher HIV stigma and discrimination dropping out of the study, thus decreasing the aggregate severity over time.

The finding that the odds of HIV viral suppression was lower in the most urban areas compared with the least urban areas, despite higher odds for utilization of care and even when controlling for health insurance, annual household income, race/ethnicity, ART adherence, and internalized HIV stigma, was alarming, and invites further exploration into the mechanisms behind the impact of population density on HIV management. This was supported by Eastwood and colleagues who found that several difficulties for participants living with HIV were reported as affecting urban participants a great deal, and more than for rural participants, such as (a) feelings of the need to “[get] back on track on your own,” (b) being “afraid nothing would help,” and (c) that they “should learn to live with it,” in addition to (d) practical issues such as transportation problems and uncertainty about housing/finances.²⁶ These findings suggest that researchers should elucidate the specific factors underlying the identified impacts of population density on health outcomes, and that clinicians should closely work with patients living in high population density areas, to prevent viral resistance and to identify STIs for treatment.

Footnotes

Acknowledgments

Data in this manuscript were collected by the Women’s Interagency HIV Study (WIHS), now the MACS/WIHS Combined Cohort Study (MWCCS). The contents of this publication are solely the responsibility of the authors and do not represent the official views of the National Institutes of Health (NIH). MWCCS (Principal Investigators): Atlanta CRS (Ighovwerha Ofotokun, Anandi Sheth, and Gina Wingood); Baltimore CRS (Todd Brown and Joseph Margolick); Bronx CRS (Kathryn Anastos, David Hanna, and Anjali Sharma); Brooklyn CRS (Deborah Gustafson and Tracey Wilson); Data Analysis and Coordination Center (Gypsyamber D’Souza, Stephen Gange and Elizabeth Topper); Chicago-Cook County CRS (Mardge Cohen and Audrey French); Chicago-Northwestern CRS (Steven Wolinsky, Frank Palella, and Valentina Stosor); Northern California CRS (Bradley Aouizerat, Jennifer Price, and Phyllis Tien); Los Angeles CRS (Roger Detels and Matthew Mimiaga); Metropolitan Washington CRS (Seble Kassaye and Daniel Merenstein; Miami CRS (Maria Alcaide, Margaret Fischl, and Deborah Jones); Pittsburgh CRS (Jeremy Martinson and Charles Rinaldo); UAB-MS CRS (Mirjam-Colette Kempf, Jodie Dionne-Odom, James B. Brock, and Deborah Konkle-Parker); UNC CRS (M. Bradley Drummond and Michelle Floris-Moore). The authors gratefully acknowledge the contributions of the study participants and dedication of the staff at the MWCCS sites.

Authors’ Contributions

D.K.P.: Conceptualization, Methodology, Funding Acquisition, Writing—Original Draft. J.D.C.: Methodology, Formal analysis, Writing—Review & Editing. D.L.: Methodology, Formal analysis, Writing—Review & Editing, Supervision. V.N.: Writing—Original Draft; Review & Editing. M.F.: Writing—Review & Editing, Funding Acquisition. G.W.: Writing—Review & Editing, Funding Acquisition. A.E.: Methodology, Writing—Review & Editing, Formal analysis.

Authors Disclosure Statement

The authors have no relevant financial or non-financial interests to disclose.

Funding Information

The MWCCS is funded primarily by the National Heart, Lung, and Blood Institute (NHLBI), with additional co-funding from the Eunice Kennedy Shriver National Institute Of Child Health & Human Development (NICHD), National Institute On Aging (NIA), National Institute Of Dental & Craniofacial Research (NIDCR), National Institute Of Allergy And Infectious Diseases (NIAID), National Institute Of Neurological Disorders And Stroke (NINDS), National Institute Of Mental Health (NIMH), National Institute On Drug Abuse (NIDA), National Institute Of Nursing Research (NINR), National Cancer Institute (NCI), National Institute on Alcohol Abuse and Alcoholism (NIAAA), National Institute on Deafness and Other Communication Disorders (NIDCD), National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), National Institute on Minority Health and Health Disparities (NIMHD), and in coordination and alignment with the research priorities of the National Institutes of Health, Office of AIDS Research (OAR): U01-HL146241, U01-HL146201, U01-HL146204, U01-HL146202, U01-HL146193, U01-HL146245, U01-HL146240, U01-HL146242, U01-HL146333, U01-HL146205, U01-HL146203, U01-HL146208, U01-HL146192, U01-HL146194. MWCCS data collection is also supported by UL1-TR000004 (UCSF CTSA), UL1-TR003098 (JHU ICTR), UL1-TR001881 (UCLA CTSI), P30-AI-050409 (Atlanta CFAR), P30-AI-073961 (Miami CFAR), P30-AI-050410 (UNC CFAR), P30-AI-027767 (UAB CFAR), P30-MH-116867 (Miami CHARM), UL1-TR001409 (DC CTSA), KL2-TR001432 (DC CTSA), and TL1-TR001431 (DC CTSA).

References

Reif

, Geonnotti

, Whetten

. HIV infection and AIDS in the Deep South. Am J Public Health, 2006; 96(6):970–973; doi: 10.2105/AJPH.2005.063149

Reif

, Pence

, Hall

, et al. HIV diagnoses, prevalence and outcomes in nine southern states. J Community Health, 2015; 40(4):642–651; doi: 10.1007/s10900-014-9979-7

Reif

, Safley

, McAllaster

, et al. State of HIV in the US Deep South. J Community Health, 2017; 42(5):844–853; doi: 10.1007/s10900-017-0325-8

Centers for Disease Control and Prevention. Monitoring selected national HIV prevention and care objectives by using HIV surveillance data - United States and 6 dependent areas, 2020. HIV Surveillance Report, 2022; 27(3).

Sullivan

, Woodyatt

, Koski

, et al. A data visualization and dissemination resource to support HIV prevention and care at the local level: Analysis and uses of the AIDSVu public data resource. J Med Internet Res, 2020; 22(10):e23173; doi: 10.2196/23173i

AIDSVu. Available from: https://map.aidsvu.org/map [Last accessed: January, 05 ].

Phillips

, Moneyham

, Thomas

, et al. Social context of rural women with HIV/AIDS. Issues Ment Health Nurs, 2011; 32(6):374–381; doi: 10.3109/01612840.2011.568273

Uphold

, Mkanta

. The use of health care services among persons living with HIV infection: State of the science and future directions. AIDS Patient Care STDS, 2005; 19(8):473–485; doi: 10.1089/apc.2005.19.473

Kalichman

, Katner

, Banas

, et al. Population density and AIDS-related stigma in large-urban, small-urban, and rural communities of the southeastern USA. Prev Sci, 2017; 18(5):517–525; doi: 10.1007/s11121-017-0761-9

10.

Vyavaharkar

, Moneyham

, Tavakoli

, et al. Social support, coping, and medication adherence among HIV-positive women with depression living in rural areas of the southeastern United States. AIDS Patient Care STDS, 2007; 21(9):667–680; doi: 10.1089/apc.2006.0131

11.

Kempf

M-C

, McLeod

, Boehme

, et al. A qualitative study of the barriers and facilitators to retention-in-care among HIV-positive women in the rural Southeastern United States: Implications for targeted interventions. AIDS Patient Care STDS, 2010; 24(8):515–520; doi: 10.1089/apc.2010.0065

12.

Kimmel

, Masiano

, Bono

, et al. Structural barriers to comprehensive, coordinated HIV care: Geographic accessibility in the US South. AIDS Care, 2018; 30(11):1459–1468; doi: 10.1080/09540121.2018.1476656

13.

Klein

, Geiger

, Chavis

, et al. The health resources and services administration’s Ryan White HIV/AIDS program in rural areas of the United States: Geographic distribution, provider characteristics, and clinical outcomes. PLoS One, 2020; 15(3):e0230121; doi: 10.1371/journal.pone.0230121

14.

Bono

, Dahman

, Sabik

, et al. Human immunodeficiency virus-experienced clinician workforce capacity: Urban-rural disparities in the southern United States. Clin Infect Dis, 2021; 72(9):1615–1622; doi: 10.1093/cid/ciaa300

15.

Reif

, Whetten

, Ostermann

, et al. Characteristics of HIV-infected adults in the Deep South and their utilization of mental health services: A rural vs. urban comparison. AIDS Care, 2006; 18(Suppl 1):S10–7; doi: 10.1080/09540120600838738

16.

Konkle-Parker

, Amico

, Henderson

. Barriers and facilitators to engagement in HIV clinical care in the Deep South: Results from Semi-structured patient interviews. J Assoc Nurses AIDS Care, 2011; 22(2):90–99; doi: 10.1016/j.jana.2010.06.002

17.

Pathman

, Fowler-Brown

, Corbie-Smith

. Differences in access to outpatient medical care for black and white adults in the rural South. Med Care, 2006; 44(5):429–438; doi: 10.1097/01.mlr.0000207487.85689.a8

18.

Nelson

, Kinder

, Johnson

, et al. Differences in selected HIV care continuum outcomes among people residing in rural, urban, and metropolitan areas-28 US jurisdictions. J Rural Health, 2018; 34(1):63–70; doi: 10.1111/jrh.12208

19.

Ohl

, Perencevich

, McInnes

, et al. Antiretroviral adherence among rural compared to urban veterans with HIV infection in the United States. AIDS Behav, 2013; 17(1):174–180; doi: 10.1007/s10461-012-0325-8

20.

Ohl

, Tate

, Duggal

, et al. Rural residence is associated with delayed care entry and increased mortality among veterans with human immunodeficiency virus infection. Med Care, 2010; 48(12):1064–1070; doi: 10.1097/MLR.0b013e3181ef60c2

21.

Sheehan

, Fennie

, Mauck

, et al. Retention in HIV care and viral suppression: Individual- and neighborhood-level predictors of racial/ethnic differences, Florida, 2015. AIDS Patient Care STDS, 2017; 31(4):167–175; doi: 10.1089/apc.2016.0197

22.

Edmonds

, Haley

, Tong

, et al. Associations between population density and clinical and sociodemographic factors in women living with HIV in the Southern United States. AIDS Care, 2021; 33(2):229–238; doi: 10.1080/09540121.2020.1769829

23.

Gonzalez

, Miller

, Solomon

, et al. Size matters: Community size, HIV stigma, & gender differences. AIDS Behav, 2009; 13(6):1205–1212; doi: 10.1007/s10461-008-9465-2

24.

Nwangwu-Ike

, Saduvala

, Watson

, et al. HIV diagnoses and viral suppression among US women in rural and nonrural areas, 2010-2017. Journal of Rural Health, 2020; 36(2):217–223; doi: 10.1111/jrh.12384

25.

Yabroff

, Lawrence

, King

, et al. Geographic disparities in cervical cancer mortality: What are the roles of risk factor prevalence, screening, and use of recommended treatment? J Rural Health, 2005; 21(2):149–157; doi: 10.1111/j.1748-0361.2005.tb00075.x

26.

Eastwood

, Fletcher

, Quinlivan

, et al. Baseline social characteristics and barriers to care from a special projects of national significance women of color with HIV study: A comparison of urban and rural women and barriers to HIV care. AIDS Patient Care STDS, 2015; 29(Suppl 1):S4–10; doi: 10.1089/apc.2014.0274

27.

Adimora

, Ramirez

, Benning

, et al. Cohort profile: The Women’s interagency HIV study (WIHS). Int J Epidemiol, 2018; 47(2):393–394i; doi: 10.1093/ije/dyy021

28.

Bacon

, von Wyl

, Alden

, et al. The Women’s Interagency HIV Study: An observational cohort brings clinical sciences to the bench. Clin Diagn Lab Immunol, 2005; 12(9):1013–1019; doi: 10.1128/CDLI.12.9.1013-1019.2005

29.

Barkan

, Melnick

, Preston-Martin

, et al. The women’s interagency HIV study. Epidemiology, 1998; 9(2):117–125.

30.

D’Souza

, Bhondoekhan

, Benning

, et al. Characteristics of the MACS-WIHS combined cohort Study: Opportunities for research on aging with HIV in the longest US observational study of HIV. Am J Epidemiol, 2021; 190(8):1457–1475; doi: 10.1093/aje/kwab050

31.

U.S. Census Bureau. American Community Survey (ACS). Available from: https://www.census.gov/programs-surveys/acs [Last accessed: commen , t].

32.

Sherbourne

, Stewart

. The MOS social support survey. Soc Sci Med, 1991; 32(6):705–714; doi: 10.1016/0277-9536(91)90150-b

33.

Bunn

, Solomon

, Miller

, et al. Measurement of stigma in people with HIV: A Reexamination of the HIV Stigma Scale. AIDS Educ Prev, 2007; 19(3):198–208; doi: 10.1521/aeap.2007.19.3.198

34.

Heckman

, Somlai

, Kalichman

, et al. Psychosocial differences between urban and rural people living with HIV/AIDS. J Rural Health, 1998; 14(2):138–145; doi: 10.1111/j.1748-0361.1998.tb00615.x

35.

Heckman

, Anderson

, Sikkema

, et al. Emotional distress in nonmetropolitan persons living with HIV disease enrolled in a telephone-delivered, coping improvement group intervention. Health Psychol, 2004; 23(1):94–100; doi: 10.1037/0278-6133.23.1.94

36.

U.S. Census Bureau. Urban and Rural. 2023 updated March 1, 2023. Available from: https://www.census.gov/programs-surveys/geography/guidance/geo-areas/urban-rural.html2023