Abstract
Objective
People who inject drugs are among the groups most vulnerable to HIV infection. The objective of this study was to describe differences in the geographic distribution of HIV diagnoses and social determinants of health (SDH) among people who inject drugs (PWID) who received an HIV diagnosis in 2017.
Methods
We used data from the National HIV Surveillance System (NHSS) to determine the counts and percentages of PWID aged ≥18 with HIV diagnosed in 2017. We combined these data with data from the US Census Bureau’s American Community Survey at the census tract level to examine regional, racial/ethnic, and population-area-of-residence differences in poverty status, education level, income level, employment status, and health insurance coverage.
Results
We observed patterns of disparity in HIV diagnosis counts and SDH among the 2666 PWID with a residential address linked to a census tract, such that counts of HIV diagnosis increased as SDH outcomes became worse. The greatest proportion of PWID lived in census tracts where ≥19% of the residents lived below the federal poverty level, ≥18% of the residents had <high school diploma, the median annual household income was <$40 000, and ≥16% of the residents did not have health insurance or a health coverage plan.
Conclusion
To our knowledge, our study is the first large-scale, census tract–level study to describe SDH among PWID with diagnosed HIV in the United States. The findings of substantial disparities in SDH among people with HIV infection attributed to injection drug use should be further examined. Understanding the SDH among PWID is crucial to reducing disparities in HIV diagnoses in this population.
The risk of acquiring HIV among people who inject drugs (PWID) is estimated to be 22 times higher than among people who do not inject drugs. 1 Disparities in health among PWID contribute to substantial population-level health disparities. 2 -4 In 2017, of the 3585 diagnosed HIV infections attributed to injection drug use (IDU; including male-to-male sexual contact and IDU), 45% were among non-Hispanic White adults and adolescents, 29% among non-Hispanic Black/African American (hereinafter referred to as non-Hispanic Black) adults and adolescents, and 20% among Hispanic/Latino adults and adolescents in the United States. 5
IDU is shaped by contextual factors, or the conditions in which people live, work, and play; these factors, which can affect health outcomes, are commonly known as social determinants of health (SDH). 6 Risk factors for HIV infection among PWID include both social determinants and behavioral determinants. In 2017 in the United States and Puerto Rico, the largest percentages of diagnosed HIV among both sexes were among people who lived in census tracts in which ≥19% of residents lived below the federal poverty level (FPL), ≥18% had less than a high school diploma, the median annual household income was <$40 000, and ≥16% did not have health insurance. 7 Poorer SDH outcomes among people diagnosed with HIV, as compared with people without HIV, is universal, regardless of HIV transmission category. However, PWID are highly vulnerable to SDH inequalities, such as homelessness, food insecurity, lack of social support, and poor access to health care. 8
The opioid epidemic has led to an increase in the number of PWID, placing new populations at risk for HIV. Although the rates of HIV diagnosis in nonurban areas have been historically low, the opioid epidemic has resulted in outbreaks of HIV among PWID in these areas. One such outbreak occurred in Scott County, Indiana, where 181 new cases of HIV were identified from November 2014 to October 2015. 9 More recently, the number of HIV diagnoses among PWID increased in nonrural cities such as Lowell and Lawrence, Massachusetts: 52 new HIV cases in 2017 were reported in the northeast region of the state among PWID, compared with 23 in 2016. 10 Another such example is that of the rapid HIV transmission between networks of PWID in northern Kentucky and the Cincinnati metropolitan region during 2017-2018. This area had fewer than 20 annual cases in previous years, but 134 PWID received an HIV diagnosis from January 2017 to October 2018. 11
The objective of our study, an ecological analysis, was to describe differences in the geographic distribution of HIV diagnoses and SDH among PWID who received an HIV diagnosis in 2017. This information will help further our understanding of how HIV diagnoses are distributed and determine areas where prevention efforts should be focused. These findings can inform effective prevention planning and resource allocation to reduce HIV transmission among PWID.
Methods
We used data from 2 sources: the Centers for Disease Control and Prevention’s (CDC’s) National HIV Surveillance System (NHSS) 12 and the US Census Bureau’s American Community Survey (ACS) 2013-2017 five-year estimates. 13 We did not seek institutional review board approval for this study because our research was a public health activity and did not involve human subjects.
National HIV Surveillance System
We based the counts and percentages of diagnoses of HIV infection on cases reported to NHSS through December 2018 and included cases among PWID aged ≥18 whose HIV infection was diagnosed during 2017. We geocoded the addresses of residence at the time of the diagnosis to the census tract level and linked these data to SDH indicator variable data from the ACS. The category PWID comprises 2 subcategories of mode of transmission: (1) PWID whose IDU was the mode of HIV transmission (category termed “IDU” hereinafter) and (2) men who reported sex with a man (MSM) and whose IDU was the mode of HIV transmission (category termed MSM/IDU hereinafter). Because a substantial proportion of cases of HIV infection are reported to CDC without an identified risk factor, we used multiple imputation to assign a transmission category. 14,15 Multiple imputation is a statistical approach in which each missing transmission category is replaced with a set of plausible values that represent the uncertainty about the true, but missing, value. 15 Each resulting data set containing the plausible values is analyzed by using standard procedures, and the results from these analyses are then combined to produce the final results. The NHSS is a database of all people in the United States and 6 dependent areas with diagnosed HIV infection. The completeness of reporting of HIV infection to NHSS is estimated to be 85%. 5 Because NHSS is a population-based census that contains data on nearly all people with diagnosed HIV in the United States and 6 dependent areas and is a comprehensive national surveillance system, we did not apply tests of significance.
American Community Survey
We obtained census tract–level SDH data from the ACS 2013-2017 five-year estimates. 13 We examined 5 SDH variables: poverty status, education level, median annual household income, employment status, and health insurance coverage. We categorized SDH indicator variables by using empirically derived quartiles. We determined quartile cut points by using data from all census tracts in the United States (Box).
Quartile cut points for variables on social determinants of health, American Community Survey, 2013-2017 5-year estimates a
aData source: US Census Bureau. 13
bQuartiles empirically derived and rounded to the nearest integer.
We used SAS version 9.4 (SAS Institute, Inc) to analyze the distribution of HIV diagnoses by selected characteristics (age group, sex at birth, race/ethnicity, transmission category, population area of residence [PAR] at diagnosis) and SDH among adults with diagnosed HIV infection attributed to IDU, stratified by census region and PAR. Because of the lack of population data on PWID, we did not calculate rates. We excluded cases or census tracts if (1) the address was nonresidential (eg, military base, corrections facility), (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 the ACS. Reported numbers <12 and their accompanying percentages should be interpreted with caution and because of their unstable rates. We did not analyze data among the Hispanic/Latino population because of small cell sizes in SDH subcategories. Regional data were based on the 4 US Census regions: Northeast, South, Midwest, and West. We defined PAR as urban (metropolitan area, ≥500 000 population), suburban (metropolitan area, 50 000-499 999 population), or rural (nonmetropolitan population, <50 000 population). 16
Results
Among PWID in the United States, 3585 HIV infections were diagnosed in 2017. Of these, 2666 (74.4%) case records included residential address information that could be linked to a census tract; 1906 (71.5%) of these were among men (Table 1).
Abbreviations: —, does not apply; AI/AN, American Indian/Alaska Native; IDU, injection drug use; MSM, men who have sex with men; NH/OPI, Native Hawaiian/Other Pacific Islander.
aNortheast: Connecticut, Maine, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, and Rhode Island; Midwest: Illinois, Indiana, Iowa, Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin; South: Alabama, Arkansas, Delaware, District of Columbia, Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and West Virginia; West: Alaska, Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming.
cNumerators were imputed and rounded to integers; percentages may not add to 100 because of rounding. Numbers <12 and percentages based on these numbers should be interpreted with caution.
dData have been statistically adjusted to account for missing data on transmission.
eData on denominators not available.
fUrban: population ≥500 000; suburban: population 50 000-499 999; rural: nonmetropolitan area, population <50 000.
gDoes not represent all adults whose HIV infection was diagnosed during 2017 in these areas, because we did not have information at the census tract level on all HIV diagnoses.
Overall, the largest proportions of PWID with diagnosed HIV were living in urban areas (men, 77.9%; women, 79.8%) at the time of diagnosis. By region, people aged 25-34 made up the highest proportion (men, 35.7%; women, 30.2%) among all age groups in the study. In all regions except the Northeast, where most diagnoses were among non-Hispanic Black adults (men, 36.7%; women, 40.5%), diagnoses occurred most often among non-Hispanic White men (44.2%) and non-Hispanic White women (45.1%). Among men overall, approximately 52.7% of diagnoses were attributed to MSM/IDU, with the exception of the Northeast, where 68.5% of diagnoses were attributed to IDU only. Among women, by region, the largest number and percentage of PWID with diagnosed HIV were in the South (42.9%), followed by the Northeast (22.8%) (Table 1).
By SDH regardless of sex, the percentages of adults with diagnosed HIV attributed to IDU were highest among adults who lived in census tracts with the least favorable outcomes: where ≥19% of residents lived below the FPL (men, 39.4%; women, 46.7%), ≥18% of residents had <high school diploma (men, 39.0%; women, 44.5%), the median annual household income was <$40 000 (men, 35.5%; women, 43.1%), ≥6% of residents were unemployed (men, 30.9%; women, 36.7%), and ≥16% of residents did not have health insurance or a health coverage plan (men, 39.8%; women, 42.5%) (Table 1).
By race/ethnicity and region, among non-Hispanic Black men and women with HIV attributed to IDU, the largest percentages of diagnoses were among people who lived in census tracts with the least favorable SDH outcome in every category and every region, except unemployment in the West and health insurance coverage in the Northeast and Midwest (Table 2). Overall, by region, although a greater percentage of non-Hispanic White men resided in census tracts with the least favorable outcome in each SDH category except unemployment than in census tracts with better outcomes, the distribution of SDH outcomes among non-Hispanic White men was less homogeneous than it was among non-Hispanic Black men. Among non-Hispanic White women in every region, more women lived in census tracts with the least favorable outcome in poverty, education, and annual household income than in census tracts with better outcomes.
Abbreviations: IDU, injection drug use; MSM, men who have sex with men.
aNortheast: Connecticut, Maine, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, and Rhode Island; Midwest: Illinois, Indiana, Iowa, Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin; South: Alabama, Arkansas, Delaware, District of Columbia, Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and West Virginia; West: Alaska, Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming.
cNumerators were imputed and rounded to integers; percentages may not add to 100 because of rounding. Numbers <12 and percentages based on these numbers should be interpreted with caution.
dData have been statistically adjusted to account for missing data on transmission.
eUrban: population ≥500 000; suburban: population 50 000-499 999; rural: nonmetropolitan area, population <50 000.
fDoes not represent all adults whose HIV infection was diagnosed during 2017 in these areas, because we did not have information at the census tract level on all HIV diagnoses.
By transmission category, among men with HIV infection attributed to IDU and MSM/IDU, the largest percentages of diagnoses were among men aged 25-34, regardless of region (except for men who inject drugs in the Northeast, where the largest percentage was among men aged 45-54) (Table 3). By race/ethnicity, the largest percentages were among non-Hispanic White men regardless of region (except for IDU in the Northeast, where the largest percentage was among non-Hispanic Black men). By population, the largest percentages were among men residing in urban areas, regardless of region. By SDH, men who injected drugs and who lived in census tracts with the least favorable outcome in poverty, education, annual household income (except the West), and unemployment (except the South) made up the largest percentage of diagnoses. By SDH among MSM who inject drugs, we observed similar overall patterns in each region, except the Northeast. In the Northeast, of the MSM who inject drugs and have sex with men, 72.8% (n = 84) lived in a census tract in the 3 lower quartiles of poverty, 30.0% (n = 35) in the second highest quartile of education, 58.6% (n = 68) in the 2 upper quartiles of annual household income, and 38.0% (n = 44) in the highest category of health insurance coverage.
Abbreviations: AI/AN, American Indian/Alaska Native; IDU, injection drug use; MSM, men who have sex with men; NH/OPI, Native Hawaiian/Other Pacific Islander.
aData have been statistically adjusted to account for missing data on transmission.
bNortheast: Connecticut, Maine, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, and Rhode Island; Midwest: Illinois, Indiana, Iowa, Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin; South: Alabama, Arkansas, Delaware, District of Columbia, Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and West Virginia; West: Alaska, Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming.
dNumerators were imputed and rounded to integers; percentages may not add to 100 because of rounding. Numbers <12 and percentages based on these numbers should be interpreted with caution.
eUrban: population ≥500 000; suburban: population 50 000-499 999; rural: nonmetropolitan area, population <50 000.
fDoes not represent all adults whose HIV infection was diagnosed during 2017 in these areas, because we did not have information at the census tract level on all HIV diagnoses.
By PAR and region, regardless of sex, the percentages of men and women diagnosed with HIV attributed to IDU were highest in urban areas (Table 4). In urban areas, the largest percentages were among people who lived in census tracts where ≥19% of residents lived below the FPL, ≥18% of residents had <high school diploma, the median annual household income was <$40 000 (except for men in the West), and ≥16% of residents did not have health insurance. Although most suburban and rural areas had small cell sizes in each SDH subcategory, the urban, suburban, and rural areas in the South had similar patterns.
Abbreviations: AI/AN, American Indian/Alaska Native; IDU, injection drug use; MSM, men who have sex with men; NH/OPI, Native Hawaiian/Other Pacific Islander.
aUrban: population ≥500 000; suburban: population 50 000-499 999; rural: nonmetropolitan area, population <50 000.
bNortheast: Connecticut, Maine, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, and Rhode Island; Midwest: Illinois, Indiana, Iowa, Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin; South: Alabama, Arkansas, Delaware, District of Columbia, Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and West Virginia; West: Alaska, Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming.
dNumerators were imputed and rounded to integers; percentages may not add to 100 because of rounding. Numbers <12 and percentages based on these numbers should be interpreted with caution.
eData have been statistically adjusted to account for missing data on transmission.
fDoes not represent all adults whose HIV infection was diagnosed during 2017 in these areas, because we did not have information at the census tract level on all HIV diagnoses.
Discussion
To our knowledge, our analysis is the first large-scale, census tract–level study to examine SDH by region and PAR among PWID with diagnosed HIV in the United States. Among PWID with diagnosed HIV and regardless of sex, non-Hispanic White adults comprised the largest proportions in every region except the Northeast. Previous studies showed that non-Hispanic White adults historically have had a higher incidence of IDU than adults of other races 17,18 and that HIV diagnoses have increased among non-Hispanic White adults who inject drugs. 18 The percentage of HIV diagnoses among people who injected drugs for the first time during the 5 years before being interviewed as part of the National HIV Behavioral Surveillance system was higher in 2014 among Non-Hispanic White people than among any other racial/ethnic population in the United States. 17 In addition, rural areas, which have predominantly non-Hispanic White populations, encompass some of the most vulnerable populations for IDU and HIV outbreaks. 16 These areas may have the greatest unmet need for syringe services programs. 19,20
Our study also showed that the racial/ethnic disparity among PWID is the most noticeable among men who reside in the Northeast, where non-Hispanic Black men tend to live in census tracts with the worst SDH outcomes and non-Hispanic White men live in census tracts with better outcomes. The distribution of SDH variables among MSM who inject drugs in the Northeast also differs from the distribution of SDH variables in other regions. Compared with MSM who inject drugs in other regions and men who inject drugs in the Northeast, MSM who inject drugs in the Northeast lived in census tracts with lower levels of poverty, higher levels of education, higher median annual household income, and smaller percentages of people without health insurance.
Our study showed that the highest percentage of PWID with diagnosed HIV lived in urban areas, regardless of region. Although outbreaks of injection drug–related HIV have mostly been observed in rural and suburban areas, it is still largely an urban problem. 10,11,21 Our findings correspond with the observations of recent outbreaks of HIV infection among PWID in Lawrence and Lowell, Massachusetts, and the Cincinnati metropolitan area. These outbreaks, primarily among PWID, occurred mostly in urban areas with long-standing opioid-related problems. 10,11 Opioid use is a problem with far‐reaching implications for the health, social, and economic well-being of all populations; in 2018 alone, approximately 10.3 million people abused opioids in the United States. 22
Our results support the literature that certain social and economic factors, location of treatment services, and stigma can shape the risk behavior of PWID. 23 -25 Poverty, for example, may reduce access to testing, treatment, and syringe exchange programs and lead to needle-sharing behaviors. Economic marginalization may lead to behavior associated with an increased risk of HIV, such as sex work or selling sex for drugs. 26,27 Previous studies showed that the health risks of PWID are inextricably bound to the social environment and, as a result, people living in environments with poor SDH are more likely to have poorer health outcomes than people living in environments with good SDH. 28 -30 Difficult political environments and lack of treatment services likely contribute to a greater number of PWID. 23 Location of treatment services and transportation difficulties also prevent PWID from receiving clean needles and other supplies. 24 PWID also experience stigma in various forms in health services contexts and from the general public, which contribute to adverse health outcomes. 25,31 PWID in our study lived in census tracts with the lowest levels of formal education. We also theorize that lack of public health education exists among PWID living in environments with poor SDH. Previous studies showed that a lack of awareness or education about safe injection is a major reason for sharing needles, which puts people at increased risk of HIV infection. 32,33
Limitations
Our analysis had several limitations. First, diagnoses of HIV infection do not represent incidence or new infections. The time from infection to diagnosis varies by individual, and residence at HIV diagnosis may not be the residence at the time HIV infection was acquired. Second, multiple imputation was used to replace missing values on transmission with a set of plausible values that represent the uncertainty about the true, but missing, value. For our study, 770 (28.4%) cases contained an imputed transmission category. Third, data were limited to people whose residential addresses were complete and, thus, could be geocoded; therefore, results may not reflect the entire population of PWID with diagnosed HIV in those areas. However, 74.4% of PWID in NHSS with HIV diagnosed in 2017 had residential addresses that were geocoded. Fourth, given that SDH information is not available at the individual level and our study was an ecological analysis, we used census tract data as a surrogate to represent the environment in which people with diagnosed HIV infection lived at the time of diagnosis. Therefore, the conclusions and findings should be interpreted with caution and not inferred at the individual level. Fifth, when stratifying the data by selected demographic and SDH categories, several subcategories had small cell sizes. Lastly, we were unable to calculate and compare diagnosis rates among PWID because of lack of denominator data for the population.
Conclusion
PWID continue to be disproportionately affected by HIV. Examining and understanding SDH factors among PWID are crucial in preventing and reducing HIV diagnoses in the United States, particularly in rural areas. Our findings show that low educational attainment, poverty, and low income were common themes among PWID with HIV. Additional work is needed to target HIV prevention interventions, including increased testing and care, for PWID in areas with high rates of poverty, low levels of education, and low income for maximum impact. The federal initiative, Ending the HIV Epidemic: A Plan for America, 34 cited PWID as a priority population. One primary strategy of the initiative is to use proven interventions to prevent new HIV transmissions. A proven intervention for preventing HIV transmission among PWID is the implementation of syringe services programs in communities that are threatened by the opioid epidemic and observing a rise in IDU. As federal agencies work to increase access to and use of comprehensive syringe services programs, consideration should be given to distributing these programs among PWID with HIV, particularly those living in areas with poor SDH outcomes.
Footnotes
Acknowledgments
The authors thank the state and territorial health departments and the HIV surveillance programs for providing their data to the Centers for Disease Control and Prevention (CDC). We also thank our CDC colleagues for their review and feedback on this article. The findings and conclusions in this article are those of the authors and do not necessarily represent the official position of CDC.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
