Abstract
Objective
To characterize migration patterns among people diagnosed as having and who died of acquired immunodeficiency syndrome (AIDS) from 1993 to 2007 because migrating to a new community can disrupt human immunodeficiency virus/AIDS care delivery and patients’ adherence to care and affect migrants’ social services and healthcare needs.
Methods
Florida AIDS surveillance data were used to describe patterns of migration among people diagnosed as having and who died of AIDS from 1993 to 2007. Individual and community characteristics were compared between residence at the time of AIDS diagnosis and residence at the time of death by type of migration.
Results
Of 31,816 people in the cohort, 2510 (7.9%) migrated to another county in Florida and 1306 (4.1%) migrated to another state. Interstate migrants were more likely to be men, 20 to 39 years old, non-Hispanic white, and born in the United States, to have had a transmission mode of injection drug use (IDU) or men who have sex with men with IDU (MSM&IDU), and to have been diagnosed before 1999. Intercounty migrants were more likely to be non-Hispanic white, younger than 60 years, have had a transmission mode of MSM, IDU, or MSM&IDU, have higher CD4 counts/percentages, and to have lived in areas with low levels of poverty or low physician density. There was a small net movement from urban to rural areas within the state.
Conclusions
A sizable percentage of people, particularly younger people and people with a transmission mode of IDU and IDU&MSM, migrated at least once between the time of their AIDS diagnosis and death. This has important implications for care and treatment, as well as efforts to prevent the disease. Further research is needed to explore barriers and facilitators to access to care upon migration and to assess the need for programs to help people transfer their human immunodeficiency virus/AIDS care, ensuring continuity of care and adherence.
Keywords: human immunodeficiency virus (HIV), acquired immunodeficiency syndrome (AIDS), substance use, intravenous, migration, access to care, rural health
Understanding patterns of migration (ie, moving to another county or farther) and characteristics of migrants with human immunodeficiency virus (HIV) infection is important because moving to a new community affects migrants and the communities to which they migrate. Although moving can result in positive outcomes such as improved access to health care and social services, moving can be stressful, may result in loss of social support, and may disrupt treatment, thus adversely affecting clinical outcomes.1–4 A net in-migration of people infected with HIV can affect the social services and healthcare systems of communities.2,3,5,6 Federal funding and resource distribution for HIV prevention and care are partially based on where people are first diagnosed and reported to local and state health departments.7 As such, any areas having a net in-migration after HIV diagnosis would be disadvantaged with respect to receipt of resources.
Previous studies indicate that people diagnosed as having HIV are relatively mobile, although longer-distance moves are less common than shorter-distance moves. The 1996 HIV Cost and Services Utilization Survey, which used a representative sample of people receiving HIV-related clinical care, indicated that 32% of respondents had moved to a different city, but only 17% had moved to a noncontiguous county or farther.8,9 Long-distance movers tend to be disproportionately men, non-Hispanic white, and have a reported HIV transmission mode of men who have sex with men (MSM).9–12 Long-distance moving also has been associated with injection drug use (IDU).9,12
The HIV/acquired immunodeficiency syndrome (AIDS) epidemic in the United States has changed over time with respect to the characteristics of people who are reported to have HIV (eg, a higher percentage of minorities and women) and to improved treatment, which has substantially lengthened survival,13–17 and these factors may affect migration patterns. One study analyzing migration patterns among a population-based sample of people with AIDS considered deaths only through 2001 and did not analyze moves within states.18
To address these information gaps, we sought to describe migration patterns between the time of diagnosis and death among people in Florida diagnosed as having AIDS, the state with the third highest HIV diagnosis rate in the United States in 2011.19 Furthermore, we sought to describe characteristics associated with migration.
Methods
Ascertainment of Cases and Survival Status
Deidentified records of Florida residents diagnosed as having AIDS and meeting the Centers for Disease Control and Prevention’s AIDS case definition20 during 1993–2007 were obtained from the Florida Department of Health (DOH) Enhanced HIV/AIDS Reporting System (eHARS). Deaths were ascertained by linking eHARS records with death certificate records from the Florida DOH Office of Vital Statistics, the Social Security Administration Death Master File, and the National Death Index.21 The endpoint for observation of death was December 31, 2007. Only people who died were included because the ZIP code of residence at the time of death, obtained from the death certificate, was known for almost all people. For people who did not die, there were incomplete data on residence after diagnosis for most cases. Limiting the analysis to people who had died is consistent with other population-based studies that have analyzed migration among people in the United States diagnosed as having HIV/AIDS.10,18
Individual-level variables available in the eHARS dataset were residential postal code (ZIP code) and county at time of AIDS diagnosis and death, month and year of AIDS diagnosis and death, country of birth, age at AIDS diagnosis, sex, race/ethnicity, HIV transmission mode, and whether diagnosis occurred at a corrections facility. All CD4 lymphocyte counts/percentages, measured within 3 months of AIDS diagnosis, were divided into quintiles. If both counts and percentage results were available, then the record was assigned to the lowest category. If there was no CD4 lymphocyte count/percentage within 3 months of the AIDS diagnosis and the person had been diagnosed as having an AIDS-defining illness at the time of diagnosis, then the person was classified as meeting the case definition in the AIDS-defining illness only/unknown CD4 lymphocyte count/percentage category.
Area-Level Poverty and Healthcare Access Variables
Area-level poverty data were obtained from the 2000 US Census and linked using the ZIP code tabulation area (ZCTA) as a result of the unavailability of 2000 US Census data by ZIP code.22 A ZCTA approximates a ZIP code and is built by aggregating the 2000 US Census blocks based on the ZIP code of addresses in these blocks. Records of diagnosed AIDS cases with missing or nonexistent ZIP codes were excluded from the analysis. Area-level poverty was defined by the percentage of households living in poverty within the ZCTA and classified into four groups based on the quartiles of poverty for all Florida ZCTAs.
Area healthcare access variables were obtained from the Health Resources and Services Administration Area Resource File.23 The average number of short-term general hospitals per 100 mi2 during the period 1995–2005 and the average number of actively practicing medical doctors and doctors of osteopathy per 100 mi2 from 1994 to 1996 were calculated for each county. The numbers of hospitals and physicians per square mile were considered because studies indicate that transportation is a barrier to obtaining health care among people living with HIV.24,25 The physician and hospital densities were dichotomized based on the median for all of the counties in Florida.
Rural-Urban Classification
Florida ZCTAs were divided into rural and urban areas using the ZIP code approximation of the Rural-Urban Categorization data codes (version 2.0, University of Washington WWAMI Rural Research Center, Seattle, http://depts.washington.edu/uwruca/ruca-data.php).26 Categorization C of the 33 Rural-Urban Categorization codes was chosen for this analysis because it classifies codes into rural and urban categories, with the rural category being restricted to small and large rural towns/cities.
Statistical Analysis
The US Census Bureau defines migration as “moves that cross jurisdictional boundaries, counties in particular”27; therefore, cases were classified into one of three migration categories: nonmigrants, people who did not move or moved within the county where they were diagnosed; intercounty migrants, people who moved to a contiguous or noncontiguous county within Florida; and interstate migrants, people who moved to another state. Individuals who migrated to a different country were excluded because there was a small number (29), and most out-of-country deaths were likely not ascertained. Furthermore, the few out-of-country deaths that were reported were not likely to be representative because migration to another country would only be ascertained in our data if the person was receiving Social Security benefits and the out-of-country death was reported to the Social Security Administration.
Associations among migration category and baseline individual-level and neighborhood-level factors were tested in the 1993–2007 cohort using the χ2 test for categorical variables and two-sample Wilcoxon rank sum tests for continuous variables. All of the variables associated with moving at P < 0.1 were included in two multilevel logistic models. In the first model the dependent variable was intercounty migration versus remaining in the original county, and this model excluded interstate migrations. In the second model the dependent variable was whether one was an interstate migrant. In this model the noninterstate group included both nonmigrants and intercounty migrants. The three levels used in the analysis were individual, ZCTA, and county. Additional models were conducted on the subset of cases diagnosed from 2002 to 2007 because this group had a shorter follow-up time and thus more similar opportunity to move. SAS Proc GLIMMIX (SAS Institute, Cary, NC) was used for modeling.
Three-year survival and length of survival were compared by migration status for the 2002–2004 diagnosis cohort only. Survival was analyzed in this cohort because it was the most recent cohort with complete 3-year follow-up. Multilevel logistic regression modeling was used to assess the association between migration and 3-year survival adjusting for baseline individual and neighborhood factors. As with the analysis of factors associated with migration, there were two models: intercounty migration versus non-migration and interstate migration versus remaining in the state. The three levels were individual, ZCTA, and county, and SAS Proc GLIMMIX was used for modeling.
Statistical analyses were performed using SAS version 9.2. The institutional review boards of the Florida DOH and Florida International University approved the study.
Results
In Florida, from 1993 to 2007, there were 79,338 people reported as having AIDS. Of these, 34,498 (43.5%) died by the end of 2007. The records for 1156 (3.4%) people receiving their diagnosis in a correctional facility were excluded because their mobility was restricted and health care is provided by the correctional facility and not the neighboring community. Of the remaining 33,342 people, 915 (2.7%) were excluded because of missing ZCTA of residence at time of diagnosis, 29 (0.09%) as a result of migrating to a different country, and 582 (1.7%) because of missing state at time of death or missing ZCTA information or county information for deaths within Florida. Of the remaining 31,816 people, 2510 (7.9%) were intercounty migrants and 1306 (4.1%) were interstate migrants (Table 1).
Table 1.
Comparison of baseline individual- and community-level characteristics and survival by migration category among people diagnosed as having AIDS who died, Florida, 1993–2007 (n = 31,816)
| Characteristics | Nonmigrants (n = 28,000), n (%)* | Moved to different county (n = 2510), n (%) | Moved to different state (n = 1306), n (%) | P† |
|---|---|---|---|---|
| Individual level | ||||
| Year of AIDS report | <0.0001 | |||
| 1993–1995 | 12,413 (87.6) | 1099 (7.8) | 662 (4.7) | |
| 1996–1998 | 6597 (87.1) | 611 (8.1) | 364 (4.8) | |
| 1999–2001 | 4259 (88.8) | 399 (8.3) | 140 (2.9) | |
| 2002–2004 | 3133 (89.2) | 280 (8.0) | 100 (2.9) | |
| 2005–2007 | 1598 (90.9) | 121 (6.9) | 40 (2.3) | |
| Sex | <0.0001 | |||
| Female | 8250 (91.0) | 592 (6.5) | 224 (2.5) | |
| Male | 19,750 (86.8) | 1918 (8.4) | 1082 (4.8) | |
| Race/ethnicity | <0.0001 | |||
| Non-Hispanic black | 15,962 (91.3) | 1144 (6.5) | 382 (2.2) | |
| Hispanic | 4362 (89.1) | 335 (6.8) | 200 (4.1) | |
| Non-Hispanic white | 7109 (81.1) | 958 (10.9) | 699 (8.0) | |
| Other | 567 (85.3) | 73 (11.0) | 25 (3.8) | |
| Age group at diagnosis, y | <0.0001 | |||
| <20 | 400 (91.3) | 31 (7.1) | 7 (1.6) | |
| 20–39 | 13,004 (85.7) | 1416 (9.3) | 759 (5.0) | |
| 40–59 | 12,239 (89.6) | 933 (6.8) | 488 (3.6) | |
| ≥60 | 2357 (92.8) | 130 (5.1) | 52 (2.1) | |
| Place of birth | <0.0001 | |||
| US | 22,272 (87.1) | 2131 (8.3) | 1183 (4.6) | |
| Not US | 5728 (91.9) | 379 (6.1) | 123 (2.0) | |
| Mode of transmission | <0.0001 | |||
| MSM | 8838 (85.0) | 976 (9.4) | 589 (5.7) | |
| IDU | 4478 (86.2) | 439 (8.5) | 280 (5.4) | |
| MSM&IDU | 926 (74.9) | 199 (16.1) | 111 (9.0) | |
| Heterosexual | 8490 (91.6) | 574 (6.2) | 204 (2.2) | |
| Other/unknown | 5268 (92.2) | 322 (5.6) | 122 (2.1) | |
| Lowest CD4 count/μL or CD4% category‡ | <0.0001 | |||
| <20 or <3% | 6770 (89.4) | 556 (7.3) | 249 (3.3) | |
| 20–53 or 3%–5% | 5335 (87.7) | 486 (8.0) | 262 (4.3) | |
| 54–110 or 6%–8% | 4541 (87.0) | 442 (8.5) | 234 (4.5) | |
| 111–161 or 9%–11% | 3558 (86.3) | 358 (8.7) | 205 (5.0) | |
| 162–199 or 12%–13% | 5295 (89.8) | 379 (6.4) | 221 (3.8) | |
| Met case definition with AIDS-defining condition only or unknown CD4 count (CD4%) | 2361 (85.5) | 272 (9.8) | 129 (4.7) | |
| 3-y survival after diagnosis (only includes 2002–2004 cohort) | <0.0001 | |||
| Survived | 483 (80.6) | 79 (13.2) | 37 (6.2) | |
| Did not survive | 2650 (90.9) | 201 (6.9) | 63 (2.2) | |
| Length of survival, mo | <0.0001 | |||
| Median (range) | 10 (0–71) | 21 (0–70) | 26 (0–60) | |
| IQR, Q1–Q3 | 1–28 | 5–37 | 11–42 | |
| ZCTA level | ||||
| Households below poverty level, % (median) | <0.0001 | |||
| 0–6.7 | 1503 (81.1) | 232 (12.5) | 119 (6.4) | |
| 6.8–10.4 | 3115 (83.9) | 380 (10.2) | 219 (5.9) | |
| 10.5–15.9 | 5989 (85.1) | 700 (9.9) | 350 (5.0) | |
| ≥16.0 | 17,393 (90.6) | 1198 (6.2) | 618 (3.2) | |
| Rural designation | <0.0001 | |||
| Rural area | 808 (75.0) | 215 (20.0) | 54 (5.0) | |
| Urban area | 27,192 (88.5) | 2295 (7.5) | 1252 (4.1) | |
| County level | ||||
| Av no. MD or DO/100 mi2 | <0.0001 | |||
| <19.744 | 633 (72.3) | 202 (23.1) | 40 (4.6) | |
| ≥19.744 | 27,367 (88.5) | 2308 (7.5) | 1266 (4.1) | |
| Av no. hospitals/100 mi2 | <0.0001 | |||
| <0.207 | 1148 (75.9) | 286 (18.9) | 79 (5.2) | |
| ≥0.207 | 26,852 (88.6) | 2224 (7.3) | 1227 (4.1) | |
AIDS: acquired immunodeficiency syndrome; DO: Doctor of Osteopathy; IDU: injection drug use; IQR, interquartile range; MSM: men who have sex with men; MD: Doctor of Medicine; ZCTA: ZIP code tabulation area.
Some percentages do not total to exactly 100% due to rounding.
P values from χ2 tests, except for length of survival, which was calculated using the Wilcoxon rank sum test.
For CD4 count/percent category there were 163 missing values (0.5% of total).
Bivariate Analysis
Non-migrants were disproportionally women, non-Hispanic black, older, foreign born, had a reported HIV transmission mode of heterosexual sex or other/unknown, lived in neighborhoods with higher percentages of poverty, and in urban areas. Nonmigrants also lived in counties with higher densities of physicians and hospitals. Survival was analyzed in the 2002–2004 cohort only, and in this cohort, the percentage of those who survived ≥3 years was significantly higher among both intercounty and interstate migrant groups than among nonmigrants, as was the median length of survival.
Multivariate Analysis of Factors Associated with Migration
Physician and hospital densities were highly correlated (Spearman correlation coefficient 0.83, P < 0.0001). To prevent multicollinearity, only physician density was included in the models. Physician density was chosen because one study found that physician availability, not hospital availability, affected the decision of people with HIV to seek health care locally in rural areas.24 In multilevel logistic regression, intercounty migrants compared with non-migrants were more likely to be non-Hispanic white as opposed to non-Hispanic black or Hispanic, younger than 60 years, have an MSM, IDU, or MSM with IDU (MSM&IDU) transmission mode, have higher CD4 levels, live in a ZCTA with the lowest level of poverty, and live in a county with a lower physician density. Interstate migrants, relative to those who remained in Florida, were more likely to be men, non-Hispanic white, aged 20 to 39 years, born in the United States, have an IDU or MSM&IDU transmission mode, and have been diagnosed before 1999 (Table 2).
Table 2.
aORs* and 95% CIs for migration among people who were diagnosed in Florida as having AIDS, 1993–2007, and who died by 2007, by migration category
| Characteristic | Migrated to different FL county (n = 2493) vs stayed in same FL county (n = 27,860) aOR* (95% CI) |
Migrated to different state (n = 1300) vs stayed in state (n = 30,353) aOR* (95% CI) |
|---|---|---|
| Individual level | ||
| Year of AIDS report | ||
| 1993–1995 | 1.05 (0.85–1.29) | 1.50 (1.08–2.09) |
| 1996–1998 | 1.22 (0.99–1.50) | 1.91 (1.36–2.67) |
| 1999–2001 | 1.23 (0.99–1.53) | 1.21 (0.84–1.73) |
| 2002–2004 | 1.17 (0.93–1.47) | 1.19 (0.82–1.74) |
| 2005–2007 | Ref | Ref |
| Sex | ||
| Female | 0.92 (0.81–1.03) | 0.70 (0.58–0.83) |
| Male | Ref | Ref |
| Race/ethnicity | ||
| Non-Hispanic black | 0.83 (0.74–0.94) | 0.37 (0.31–0.43) |
| Hispanic | 0.79 (0.68–0.93) | 0.77 (0.63–0.93) |
| Non-Hispanic white | Ref | Ref |
| Other | 1.24 (0.95–1.62) | 0.59 (0.39–0.90) |
| Age group at diagnosis, y | ||
| <20 | 2.05 (1.33–3.16) | 1.17 (0.52–2.64) |
| 20–39 | 1.71 (1.41–2.08) | 1.77 (1.31–2.37) |
| 40–59 | 1.22 (1.00–1.49) | 1.31 (0.97–1.76) |
| ≥60 | Ref | Ref |
| Place of birth | ||
| US | 0.96 (0.84–1.11) | 1.94 (1.56–2.42) |
| Not US | Ref | Ref |
| Mode of transmission | ||
| MSM | 1.29 (1.12–1.49) | 1.17 (0.96–1.44) |
| IDU | 1.35 (1.17–1.55) | 1.77 (1.46–2.16) |
| MSM&IDU | 2.43 (1.99–2.97) | 1.97 (1.50–2.58) |
| Heterosexual | Ref | Ref |
| Other/unknown | 0.90 (0.78–1.05) | 0.89 (0.70–1.13) |
| Lowest CD4 count/μL or CD4% category† | ||
| <20 or <3% | 0.72 (0.61–0.84) | 0.86(0.69–1.07) |
| 20–53 or 3%–5% | 0.79 (0.68–0.93) | 1.04 (0.83–1.29) |
| 54–110 or 6%–8% | 0.84 (0.71–0.99) | 1.02 (0.81–1.27) |
| 111–161 or 9%–11% | 0.86 (0.72–1.02) | 1.08 (0.86–1.36) |
| 162–199 or 12%–13% | Ref | Ref |
| Met case definition with AIDS-defining condition only or unknown CD4 count/CD4%) | 0.66 (0.55–0.78) | 0.86 (0.68–1.08) |
| ZCTA of residence at diagnosis | ||
| ZCTA households below poverty line, % | ||
| 0–6.7 | Ref | Ref |
| 6.8–10.4 | 0.84 (0.69–1.03) | 0.96 (0.74–1.23) |
| 10.5–15.9 | 0.87 (0.72–1.05) | 0.95 (0.74–1.21) |
| ≥16.0 | 0.61 (0.51–0.74) | 0.94 (0.74–1.19) |
| Rural designation | ||
| Rural area | 0.81 (0.55–1.19) | 1.04 (0.65–1.67) |
| Urban area | Ref | Ref |
| County of residence at diagnosis | ||
| Av total no. MD or DO/100 mi2 | ||
| <19.744 | 2.64 (1.77–3.94) | 1.18 (0.71–1.94) |
| ≥19.744 | Ref | Ref |
AIDS, acquired immunodeficiency syndrome; aOR, adjusted odds ratio; CI, confidence interval; DO, Doctor of Osteopathy; IDU, injection drug use; MD, Doctor of Medicine, MSM, men who have sex with men; ZCTA, ZIP code tabulation area.
ORs adjusted for all variables in the model. Statistically significant ORs (P < 0.05) shown in boldface type.
For CD4 count/percentage category, there were 157 in the model for moving out of county and 163 missing values for moving out of state.
The variables of age groups younger than 60, MSM&IDU transmission mode, high CD4 levels, and residence in an area with the lowest percentage of poverty remained significantly associated with intercounty migration when limiting the analysis to the 2002–2007 diagnosis cohort. The factors of non-Hispanic white race, IDU transmission mode, and low physician density were no longer significant, although the odds ratios were in the same direction. For interstate migration, the variables of 20- to 29-year-old age group, non-Hispanic white race compared with non-Hispanic black race, and being born in the United States remained significant. Male sex, IDU and MSM&IDU transmission modes, and non-Hispanic white race versus Hispanic and non-Hispanic white races versus other race/ethnicity were no longer significant, although the odds ratios were in the same direction (data not shown).
Rural-to-Urban and Urban-to-Rural Moves
Overall, there was a small net movement to rural areas among those who died in Florida (40 people, or 3.9% increase from original 1023; data not shown). When movers from urban to rural areas were compared with those who stayed in urban areas in Florida, the variables that were associated with moving to rural areas were being born in the United States (adjusted odds ratio [aOR] 2.88; 95% confidence interval [CI] 1.61–5.16), being male (aOR 1.65; 95% CI 1.11–2.45), an MSM&IDU mode compared with a heterosexual mode of transmission (aOR 1.95; 95% CI 1.07–3.55), and living in a county with a higher physician density (AOR 6.31; 95% CI 2.62–15.21; data not shown).
Multivariate Analysis of Factors Associated with Survival
In the multivariate analysis of 3-year survival in the 2002–2004 cohort, intercounty migration and interstate migration were significantly associated with survival after adjusting for individual-level and ZCTA-level factors (Table 3). County-level factors were not included because the models would not converge with the county-level variables. When those who died in the first 4 months after diagnosis were excluded from the analysis because they were likely too sick to move, the ORs decreased (aOR 1.65 [95% CI 1.21–2.25] and 2.23 [95% CI 1.42–3.49] for intercounty and interstate migration, respectively; data not shown).
Table 3.
3-year survival among people who were diagnosed in Florida as having AIDS, 2002–2004, and who died by 2007
| Characteristic | Intercounty model aOR* (95% CI) for 3-y survival among those diagnosed 2002–2004 who remained in the state† (n = 3398) | Interstate model aOR* (95% CI) for 3-y survival among those diagnosed 2002–2004 (n = 3497) |
|---|---|---|
| Individual level | ||
| Migrated‡ | ||
| Yes | 2.00 (1.49–2.69) | 3.00 (1.94–4.64) |
| No | Ref | Ref |
| Sex | ||
| Female | 0.95 (0.75–1.20) | 0.97 (0.77–1.22) |
| Male | Ref | Ref |
| Race/ethnicity | ||
| Non-Hispanic black | 1.28 (0.97–1.69) | 1.29 (0.99–1.69) |
| Hispanic | 1.12 (0.79–1.61) | 1.15 (0.81–1.61) |
| Non-Hispanic white | Ref | Ref |
| Other | 1.48 (0.85–2.58) | 1.47 (0.85–2.54) |
| Age group at diagnosis, y | ||
| <20 | 2.27 (0.59–8.73) | 1.94 (0.51–7.43) |
| 20–39 | 1.68 (1.18–2.40) | 1.76 (1.24–2.50) |
| 40–59 | 1.28 (0.91–1.81) | 1.31 (0.94–1.84) |
| ≥60 | Ref | Ref |
| Place of birth | ||
| US | 1.13 (0.87–1.48) | 1.17 (0.90–1.52) |
| Not US | Ref | Ref |
| Mode of transmission | ||
| MSM | 0.91 (0.69–1.20) | 0.87 (0.67–1.15) |
| IDU | 1.04 (0.78–1.38) | 1.03 (0.78–1.35) |
| MSM&IDU | 1.00 (0.59–1.71) | 1.11 (0.67–1.84) |
| Heterosexual | Ref | Ref |
| Other/unknown | 0.37 (0.26–0.51) | 0.36 (0.26–0.50) |
| Lowest CD4 count/μL or CD4% category‡ | ||
| <20 or <3% | 0.52 (0.36–0.73) | 0.50 (0.36–0.70) |
| 20–53 or 3%–5% | 0.66 (0.46–0.93) | 0.63 (0.45–0.89) |
| 54–110 or 6%–8% | 0.82 (0.58–1.17) | 0.77 (0.55–1.09) |
| 111–161 or 9%–11% | 1.20 (0.84–1.72) | 1.15 (0.81–1.62) |
| 162–199 or 12%–13% | Ref | Ref |
| Met case definition with AIDS-defining condition only or unknown CD4 count (CD4%) | 0.48 (0.31–0.73) | 0.47 (0.31–0.70) |
| ZCTA of residence at diagnosis | ||
| Households below poverty line, % | ||
| 0–6.7 | 1.34 (0.88–2.04) | 1.21 (0.81–1.80) |
| 6.8–10.4 | 1.27 (0.82–1.98) | 1.19 (0.78–1.81) |
| 10.5–15.9 | 1.05 (0.64–1.72) | 1.11 (0.70–1.79) |
| ≥16.0 | Ref | Ref |
| Rural designation | ||
| Urban area | 0.70 (0.45–1.10) | 0.63 (0.41–0.98) |
| Rural area | Ref | Ref |
AIDS, acquired immunodeficiency syndrome; aOR: adjusted odds ratio; CI, confidence interval; DO, Doctor of Osteopathy; IDU, injection drug use; MD, Doctor of Medicine; MSM, men who have sex with men; ZCTA, ZIP code tabulation area.
ORs adjusted for all variables in the model. Statistically significant ORs (P < 0.05) are shown in boldface type. For CD4 count/percent category there were 15 missing values in moving out of county; and 18 missing values in model for moving out of state.
Excludes those who left state.
In the intercounty model, migrants were people who moved to another Florida county, and non-migrants were people who stayed in the county. In the interstate model, migrants were people who moved to a new state and non-migrants were those who did not leave the state.
Discussion
This study has four major findings. First, a sizable percentage of people migrated at least once between the time of AIDS diagnosis and death. Second, younger age and IDU transmission mode were associated with migration. Third, there was a small net movement to rural areas. Fourth, migrants survived longer than did non-migrants.
The first main finding that 12% of the cohort migrated is not surprising given the high mobility of the US population in general. During the period 1995–2000, 46% of the US population ≥5 years old moved to a new address.28 A study analyzing state-to-state moves among people with AIDS who died between 1993 and 2001 in the United States found that 5.4% had moved from the state where they received their diagnosis,18 similar to the 4.1% found in our study. The finding that moving is common among people diagnosed as having AIDS has important implications in the planning of HIV prevention and care efforts. It suggests that capturing the current address in surveillance data and using it to determine the current prevalence of HIV and AIDS would provide more accurate information than using the address at the time of diagnosis. It is also important because several studies suggest that there are lapses in adherence to and delivery of care associated with moving.3,4,29
The second finding that younger age groups, particularly the 20- to 39-year-old group, were more likely to migrate is consistent with studies that have reported associations between migration and age.9,10,12,18 Because younger, relative to older, people may practice riskier sexual and needle-sharing behaviors,30–33 it is important for communities to use migration information in planning HIV prevention strategies. In addition, because younger people may experience a particularly large benefit from antiretroviral therapy (ie, they have many potential years of good-quality life to gain), ensuring a smooth transition to care after migration is particularly important.
That IDU transmission mode was associated with migration is consistent with findings from several clinic-based studies3,9,12,34 and from one population-based study in the southern region of the United States.18 A study of patients receiving HIV care in British Columbia, Canada, found that the association between migration and poor adherence was stronger for IDUs than for other groups of people with HIV infection.29 The authors of the study noted that poor adherence would not only affect the health status of IDUs who move but also would result in the IDUs having higher viral loads, thus being more infectious.29 This may pose a risk, through needle-sharing or unprotected sex, to the communities to which the IDUs migrate.
The third finding was that there was a small net movement from urban to rural areas. We have no information on why a given person may have decided to move. Other studies have found that people with HIV move to rural areas to live near their families and for social support.5,6,11
The fourth finding that people who migrated tended to survive longer does not support the idea that moving leads to disruption of care and problems with adherence; however, it is likely that migrants were healthier originally. This explanation is supported by the higher CD4 levels among migrants and the decline in survival differences among migrant groups when people who died within the first 4 months after diagnosis were excluded. It is also possible that migrants had better access to health care and other resources given that they migrated from communities with less poverty. In addition, there may be other unmeasured characteristics that are more prevalent among the migrants that enhanced their survival. As such, we cannot conclude that there is no disruption in care with moving. We did not have the data to analyze continuity or quality of care or adherence to care. It would be useful to examine data on quality and continuity of care, adherence, and clinical outcomes such as viral loads and opportunistic infections among migrants.
The present study is subject to several additional limitations. First, because there were data only on the ZCTA codes at the time of diagnosis and death, we would not know whether a person migrated multiple times and whether they in fact migrated to another state or county and then migrated back to the county where they received their diagnosis; therefore, it is likely that there was a higher percentage of people who migrated than we observed. This is probably less of a problem in interpreting the factors associated with migration because they did not change substantially when we limited our analysis to the 2002–2007 cohort, which had less time to move. Second, the study did not include people who were diagnosed after 2007 because of the unavailability of complete vital status data. Although we saw little evidence that the factors associated with migration were different in the most recent cohorts (people diagnosed 2002–2007) compared with the entire cohort (people diagnosed 1993–2007), it would be useful to assess migration patterns in people more recently diagnosed with HIV. Finally, we could not assess the migration of people between the time of their HIV diagnosis and that of their AIDS diagnosis.
Conclusions
We found that a sizeable percentage of people in Florida migrated after being diagnosed as having AIDS. Although migrants in general appeared to survive longer with the disease, additional research is needed to explore putative barriers and facilitators to access to care upon moving and to assess the need for programs to help people transfer their HIV care and treatment to ensure continuity of care and adherence to care. The migration of people after an HIV/AIDS diagnosis also should be considered in resource allocation with respect to HIV/AIDS care and treatment. The relevance of this issue is increasing because people are living longer with HIV/AIDS.
Key Points.
Migration was relatively common, with 7.9% migrating to another Florida county and 4.1% migrating to another state.
There was a small net movement from urban to rural areas.
Younger age and the human immunodeficiency virus transmission modes of injection drug use and injection drug use combined with men who have sex with men were associated with migration.
Further research is needed to explore barriers and facilitators to access to care upon moving and to assess the need for programs to help people transfer their human immunodeficiency virus/acquired immunodeficiency syndrome care, ensuring continuity of care and adherence to care.
Acknowledgments
The project described was supported by Award No. R01MD004002 from the National Institute on Minority Health and Health Disparities at the National Institutes of Health.
The authors thank Tracina Bush, BS, and Julia Fitz, MPH, for assistance in preparing the dataset.
Footnotes
The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute on Minority Health and Health Disparities or the National Institutes of Health.
M.J.T. has been a consultant to the Health Foundation of South Florida and served as a speaker for the Florida AIDS Education Training Program.
The other authors have no financial relationships to disclose and no conflicts of interest to report.
References
- 1.Tatum PS, Schoech D. Migration of persons with HIV disease: the search for care. AIDS Public Policy J. 1992;7:56–63. [Google Scholar]
- 2.Evans J, Baldwin W. Migration and health. Int Migr Rev. 1987;21:v-491–v-486. [PubMed] [Google Scholar]
- 3.Gill MJ, Krentz HB. Unappreciated epidemiology: the churn effect in a regional HIV care programme. Int J STD AIDS. 2009;20:540–544. doi: 10.1258/ijsa.2008.008422. [DOI] [PubMed] [Google Scholar]
- 4.Krentz HB, Worthington H, Gill MJ. Adverse health effects for individuals who move between HIV care centers. J Acquir Immune Defic Syndr. 2011;57:51–54. doi: 10.1097/QAI.0b013e318214feee. [DOI] [PubMed] [Google Scholar]
- 5.Cohn SE, Klein JD, Mohr JE, et al. The geography of AIDS: patterns of urban and rural migration. South Med J. 1994;87:599–606. doi: 10.1097/00007611-199406000-00004. [DOI] [PubMed] [Google Scholar]
- 6.Davis K, Stapleton J. Migration to rural areas by HIV patients: impact on HIV-related healthcare use. Infect Control Hosp Epidemiol. 1991;12:540–543. doi: 10.1086/646406. [DOI] [PubMed] [Google Scholar]
- 7.Johnson JA. The Ryan White HIV/AIDS Program. [Accessed November 2, 2012.];Congressional Research Service Report for Congress. http://www.fas.org/sgp/crs/misc/RL33279.pdf. Published June 1, 2011.
- 8.London AS, Wilmoth JM, Fleishman JA. Moving for care: findings from the US HIV Cost and Services Utilization Study. AIDS Care. 2004;16:858–875. doi: 10.1080/09540120412331290149. [DOI] [PubMed] [Google Scholar]
- 9.Berk ML, Schur CL, Dunbar JL, et al. Short report: migration among persons living with HIV. Soc Sci Med. 2003;57:1091–1097. doi: 10.1016/s0277-9536(02)00487-2. [DOI] [PubMed] [Google Scholar]
- 10.Buehler JW, Frey RL, Chu SY, et al. The migration of persons with AIDS: data from 12 states, 1985 to 1992. Am J Public Health. 1995;85:1552–1555. doi: 10.2105/ajph.85.11.1552. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Lansky A, Nakashima AK, Diaz T, et al. Human immunodeficiency virus infection in rural areas and small cities of the southeast: contributions of migration and behavior. J Rural Health. 2000;16:20–30. doi: 10.1111/j.1748-0361.2000.tb00433.x. [DOI] [PubMed] [Google Scholar]
- 12.Lieb S, Trepka MJ, Liberti TM, et al. HIV/AIDS patients who move to urban Florida counties following a diagnosis of HIV: predictors and implications for HIV prevention. J Urban Health. 2006;83:1158–1167. doi: 10.1007/s11524-006-9128-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Centers for Disease Control and Prevention. Epidemiology of HIV/AIDS—United States, 1981–2005. MMWR Morb Mortal Wkly Rep. 2006;55:589–592. [PubMed] [Google Scholar]
- 14.Detels R, Muñoz A, McFarlane G, et al. Effectiveness of potent antiretroviral therapy on time to AIDS and death in men with known HIV infection duration. Multicenter AIDS Cohort Study Investigators. JAMA. 1998;280:1497–1503. doi: 10.1001/jama.280.17.1497. [DOI] [PubMed] [Google Scholar]
- 15.Palella FJ, Delaney KM, Moorman AC, et al. Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. HIV Outpatient Study Investigators. N Engl J Med. 1998;338:853–860. doi: 10.1056/NEJM199803263381301. [DOI] [PubMed] [Google Scholar]
- 16.Schneider MF, Gange SJ, Williams CM, et al. Patterns of the hazard of death after AIDS through the evolution of antiretroviral therapy: 1984–2004. AIDS. 2005;19:2009–2018. doi: 10.1097/01.aids.0000189864.90053.22. [DOI] [PubMed] [Google Scholar]
- 17.Schwarcz SK, Hsu LC, Vittinghoff E, et al. Impact of protease inhibitors and other antiretroviral treatments on acquired immunodeficiency syndrome survival in San Francisco, California, 1987–1996. Am J Epidemiol. 2000;152:178–185. doi: 10.1093/aje/152.2.178. [DOI] [PubMed] [Google Scholar]
- 18.Harris NS, Dean HD, Fleming PL. Characteristics of adults and adolescents who have migrated from place of AIDS diagnosis to place of death, United States, 1993–2001. AIDS Educ Prev. 2005;17(Suppl B):39–48. doi: 10.1521/aeap.2005.17.Supplement_B.39. [DOI] [PubMed] [Google Scholar]
- 19.Centers for Disease Control and Prevention. [Accessed March 8, 2013.];Diagnoses of HIV infection in the United States and dependent areas. 2011 23 http://www.cdc.gov/hiv/surveillance/resources/reports/2011report/index.htm. Published February 2013. [Google Scholar]
- 20.1993 revised classification system for HIV infection and expanded surveillance case definition for AIDS among adolescents and adults. MMWR Recomm Rep. 1992;41(RR-17):1–19. [PubMed] [Google Scholar]
- 21.Trepka MJ, Maddox LM, Lieb S, et al. Utility of the National Death Index in ascertaining mortality in acquired immunodeficiency syndrome surveillance. Am J Epidemiol. 2011;174:90–98. doi: 10.1093/aje/kwr034. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.US Census Bureau. Census. [Accessed July 6, 2014.];ZCTA, ZIP Code Tabulation Areas technical documentation. 2000 https://www.census.gov/geo/reference/zctas.html.
- 23.Area Resource File. Rockville, MD: US Department of Health and Human Services, Health Resources and Services Administration, Bureau of Health Professions; 2007. [Google Scholar]
- 24.Schur CL, Berk ML, Dunbar JR, et al. Where to seek care: an examination of people in rural areas with HIV/AIDS. J Rural Health. 2002;18:337–347. doi: 10.1111/j.1748-0361.2002.tb00895.x. [DOI] [PubMed] [Google Scholar]
- 25.Sutton M, Anthony M, Vila C, et al. HIV testing and HIV/AIDS treatment services in rural counties in 10 southern states: service provider perspectives. J Rural Health. 2010;26:240–247. doi: 10.1111/j.1748-0361.2010.00284.x. [DOI] [PubMed] [Google Scholar]
- 26.Hart LG, Larsen EH, Lishner DM. Rural definitions for health policy and research. Am J Public Health. 2005;95:1149–1155. doi: 10.2105/AJPH.2004.042432. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.US Census Bureau. Glossary. [Accessed December 18, 2013.]; http://www.census.gov/glossary/#term_Migration.
- 28.Schachter JP. Migration by race and Hispanic origin: 1995 to 2000. US Census Bureau; [Accessed December 12, 2011.]. CENSR-13. http://www.census.gov/prod/2003pubs/censr-13.pdf. Published October 2003. [Google Scholar]
- 29.Lima V, Fernandes K, Rachlis B, et al. Migration adversely affects antiretroviral adherence in a population-based cohort of HIV/AIDS patients. Soc Sci Med. 2009;68:1044–1049. doi: 10.1016/j.socscimed.2008.12.043. [DOI] [PubMed] [Google Scholar]
- 30.Centers for Disease Control and Prevention. HIV infection and HIV-associated behaviors among injecting drug users—20 cities, United States, 2009. MMWR Morb Mortal Wkly Rep. 2012;61:133–138. [PubMed] [Google Scholar]
- 31.Adimora AA, Schoenbach VJ, Doherty IA. Concurrent sexual partnerships among men in the United States. Am J Public Health. 2007;97:2230–2237. doi: 10.2105/AJPH.2006.099069. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Adimora AA, Schoenbach VJ, Taylor EM, et al. Concurrent partnerships, nonmonogamous partners, and substance use among women in the United States. Am J Public Health. 2011;101:128–136. doi: 10.2105/AJPH.2009.174292. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Schwarcz S, Scheer S, McFarland W, et al. Prevalence of HIV infection and predictors of high-transmission sexual risk behaviors among men who have sex with men. Am J Public Health. 2007;97:1067–1075. doi: 10.2105/AJPH.2005.072249. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Wood E, Yip B, Gataric N, et al. Determinants of geographic mobility among participants in a population-based HIV/AIDS drug treatment program. Health Place. 2000;6:33–40. doi: 10.1016/s1353-8292(99)00028-3. [DOI] [PubMed] [Google Scholar]