The active duty component of the US Army is a dynamic cohort with the continuous addition and removal of service personnel; approximately 80,000 personnel are recruited annually. This population is composed of young, sexually active, racially/ethnically and geographically diverse men and women who are tested repeatedly throughout their careers for HIV infection. From 1985 to 2003, more than 1.5 million US Army personnel were tested for HIV infection. Previous reports have shown that the overall HIV incidence rate (IR) among active duty US Army personnel declined from 0.45 per 1000 person-years (PYs) in 1986 to 0.08 per 1000 PYs in 1999.1
The aim of the present study was to analyze changes in HIV IRs among white and African-American active duty US army personnel during the 18-year period from 1986 through 2003 by using a recently developed statistical methodology for time trend analyses, Joinpoint regression.2 Trend analyses provide insight into changes in HIV infection rates for those population subgroups with higher HIV IRs (ie, those who were younger, unmarried, enlisted personnel, and African American).
MATERIALS AND METHODS
Data on active duty US Army personnel were obtained from the US Army Medical Surveillance Activity (AMSA), which maintains records on demographics, occupation, hospitalizations, ambulatory medical visits, reportable diseases, and HIV testing history.3
This cohort study consisted of US Army personnel who were on active duty at any time between October 1985 (the beginning of the Department of Defense [DoD] HIV testing program) and 2003.4,5 Individuals in this study were included only if they had a minimum of 2 HIV tests. Active duty personnel undergo periodic and regular HIV testing for routine (ie, biennial testing, physical examination, deployment) and adjunct (performed in association with medical evaluations) indications.5 US Army personnel with prevalent HIV infections identified through their first HIV test were excluded from analysis. This study was restricted to white and African-American personnel only, because they comprised more than 90% of US Army active duty personnel. Behavioral risk factor information was not available.
Laboratory testing and control measures have been described in detail elsewhere.1 Briefly, blood serum samples were tested by an enzyme-linked immunosorbent assay (ELISA). All repeatedly positive samples were subsequently confirmed by Western blot (WB) analysis. An HIV-positive test was defined as a reactive ELISA followed by a confirmatory WB, with the same result replicated from a second serum specimen.
Variables available for analysis were year, gender, age, race, marital status, length of service, rank, and occupational group. Annual HIV IRs were expressed as the number of new infections per 1000 person-years (PYs) of follow-up, with the 95% confidence interval (CI) estimated using the Fisher exact formula. The seroconversion date was defined as the midpoint between the date of the last negative HIV test and the date of the first positive HIV test.
Joinpoint regression analysis was applied to evaluate in detail the changes in HIV IRs by all available variables.2,6 Joinpoint analyzes changing trends over successive segments of time and the amount of increase or decrease by choosing the best-fitting time periods at which the direction or magnitude of the trend changes significantly. The analysis begins with a minimum number of joinpoints and tests whether the addition of 1 or more joinpoints (to a maximum of 4) is statistically significant and should be added to the model. The number of significant joinpoints was found by performing permutation tests. The estimated annual percentage change (APC) in HIV IRs is reported for each time period by fitting a Poisson distribution. A negative APC indicates a decreasing trend, and a positive APC indicates an increasing trend.2
A total of 1,280 incident HIV infections were observed in 1.5 million white or African-American active duty US Army personnel with 8,366,423 PYs of follow-up. Of the total number of HIV seroconverters, 854 (67%) were African American, 1196 (93%) were male, and 554 (43%) were between 17 and 24 years old.
The overall HIV IR was 0.15 per 1000 PYs (95% CI: 0.14 to 0.16). Joinpoint regression analysis reported 2 significant periods of declining HIV IRs between 1986 and 2003. The greatest decline in HIV IRs (APC = −19.7, 95% CI: −25.5 to −13.3; P < 0.001) was observed in the first period (1986-1991), whereas in the second time period (1991-2003), only a slight decline in HIV IRs was noted (APC = −2.9, 95% CI: −5.6 to −0.2; P = 0.045).
Incident HIV Infections Among White Personnel
The overall HIV IR in white personnel was 0.07 per 1000 PYs (95% CI: 0.07 to 0.08; APC = −9.4, 95% CI: −12.2 to −6.5; P < 0.001). HIV IRs were significantly (P < 0.05) higher among health care professionals, unmarried individuals, personnel with <3 years of service, and enlisted personnel (Table 1). In addition, Joinpoint analysis revealed 2 significant time periods: the first between 1986 and 1989 (APC = −31.1, 95% CI: −46.1 to −11.9; P = 0.006) and the second between 1989 and 2003 (APC = −5.7, 95% CI: −8.9 to −2.4; P = 0.003). Figure 1 shows the annual HIV IRs among white and African-American personnel.
All stratified analyses revealed a significant decline in HIV IRs over time (APC range: −7.2 to −11.5; P < 0.05), with the exception of officer personnel (APC = −2.5, 95% CI: −8.6 to 4.1; P = 0.453). HIV IRs declined significantly over 2 time periods (1986 to 1989-1990 and 1989-1990 to 2003) among the following subgroups: male personnel, personnel with 3 or more years of service, enlisted personnel, and individuals in all occupations other than health care. A steady decline in the HIV IRs over the 18 years of follow-up was observed for married personnel (APC = −10.2, 95% CI: −14.1 to −6.2; P < 0.001) and for personnel with <3 years of service (APC = −8.4, 95% CI: −11.3 to −5.3; P < 0.001).
Incident HIV Infections Among African-American Personnel
The overall HIV IR in African-American personnel was 0.34 per 1000 PYs (95% CI: 0.32 to 0.36; APC = −6.8, 95% CI: −9.2 to −4.3; P < 0.001). HIV IRs were significantly higher among younger (17-24 years old), unmarried personnel, with <3 years of service, enlisted personnel, and health care professionals (Table 2). According to Joinpoint analysis, only 1 significant time period (between 1986 and 1991) of declining HIV IRs overall (APC = −19.4, 95% CI: −26.3 to −11.8; P < 0.001) was found.
Stratified analyses showed significant declines in HIV IRs among the following subgroups: male personnel, married personnel, individuals with less than or more than 3 years of service, enlisted personnel, and personnel in occupations other than health care. Most declining HIV IRs occurred between 1986 and 1991 through 1993 (Table 2).
Despite the observed decline among African-American personnel, increases in HIV IRs were noted among 2 subgroups: health care professionals (APC = 16.8, 95% CI: −0.09 to 37.82; P = 0.063) from 1995 to 2003 and unmarried personnel (APC = 3.9, 95% CI: −0.2 to 8.1; P = 0.059) from 1991 to 2003 (Fig. 2).
This cohort study provides a concise analysis of trends in HIV IRs among white and African-American active duty US Army personnel by using Joinpoint regression. This statistical methodology offers an advantage over simple annual IR comparisons for the purposes of tracking changes over time, providing better insight into any observed IR changes by documenting when and for what length of time changes occur.2
Results from this study document an overall dramatic decline in HIV IRs for white and African-American personnel in the early years (1986-1992) of the DoD testing program. The major findings among white personnel were the presence of significant steady declines in HIV IRs for the following subpopulations: male personnel, married personnel, personnel with less than or more than 3 years of service, enlisted personnel, and personnel in occupations other than health care. Among African-American personnel, most declining HIV IRs occurred between 1986 and 1991 through 1993; however, the only statistically significant steady decline was among personnel with a length of service <3 years.
Joinpoint analysis revealed an increase in HIV IRs for 2 African-American subpopulations; the first among health care professionals and the second among unmarried personnel. Previous studies have reported that the number of HIV infections acquired occupationally via needle stick injuries is relatively low in the United States.7,8 Therefore, occupational injuries are not likely to contribute significantly to the increasing HIV IRs among health care workers. Our a priori assumption had been that health care professionals would have a better knowledge of HIV prevention, and thus would have been shown to have a declining risk of HIV infection.
One possible reason for the increase in the HIV IRs among unmarried African-American personnel may be attributable to the significant correlation previously found between marital status and annual number of sexual partners among black men, with single men reporting a higher number of sexual partners.9
The main limitation of this study is the lack of individual behavioral risk factor information. Given the sensitivity surrounding HIV risk behaviors (sexual and drug use) among US military personnel, unique challenges are encountered in military HIV epidemiologic research. The most recent studies (1990 and 1995) of military personnel documented elevated risks of HIV infection associated with same-sex behavior, sexual activity resulting in contact with blood, sexual contact with prostitutes or injecting drug users,10 6 or more lifetime partners, 3 or more casual partners, or engaging in sex on the first day of meeting.11 A recent DoD survey12 documented a self-reported lifetime sexually transmitted infection (STI) prevalence of 20% among military personnel, with 44% reporting condom use during their last sexual encounter. Only 5% of all US Army personnel stated they had used any illicit drugs (other than marijuana). Because homosexual conduct disqualifies a person from enlisting in the US Military and is grounds for separation from service while on active duty, it is probable that men who have sex with men (MSM) have been discouraged from enlisting in the armed forces.13 These results suggest that the main mode of transmission is sexual intercourse.
In summary, this cohort study documents that African-American personnel had higher overall HIV IRs compared with white personnel and suggests that despite an overall decline in HIV IRs between 1986 and 2003, certain subpopulations among African-American personnel were observed to have increasing HIV IRs. Additional studies are currently underway that should further define risk factors for HIV infection. These data should contribute to the development of targeted preventive interventions in specific subpopulations at risk for HIV infection in the US Army.
The authors gratefully acknowledge the contribution of Mark J. Milazzo at the Henry M. Jackson Foundation for the Advancement of Military Medicine, Inc., Rockville, MD, the assistance provided by COL. Mark Rubertone at the Army Medical Surveillance Activity, Washington, DC, and the technical assistance of Andrew Sebastian.
1. Renzullo PO, Sateren WB, Garner RP, et al. HIV
-1 seroconversion in United States
Army active duty personnel, 1985-1999. AIDS
2. Kim HJ, Fay MP, Feuer EJ, et al. Permutation tests for Joinpoint
regression with applications to cancer rates. Stat Med
. 2000;19:335-351 [erratum in Stat Med
3. Rubertone MV, Brundage JF. The Defense Medical Surveillance System and the Department of Defense serum repository: glimpses of the future of public health surveillance. Am J Public Health
4. US Department of the Army. Army Regulation 600-110: Personnel-General, Identification, Surveillance, and Administration of Personnel Infected with Human Immunodeficiency Virus (HIV)
. Washington, DC: Headquarters, Department of the Army; 2005.
5. US Department of Defense. Health Policy HA 04-007, Policy Memorandum-Human Immunodeficiency Virus Interval Testing, Assistant Secretary of Defense for Health Affairs
. Washington, DC: US Department of Defense; 2004.
7. Do AN, Ciesielski CA, Metler RP, et al. Occupationally acquired human immunodeficiency virus (HIV
) infection: national case surveillance data during 20 years of the HIV
epidemic in the United States
. Infect Control Hosp Epidemiol
8. Gerberding JL. Clinical practice. Occupational exposure to HIV
in health care settings. N Engl J Med
9. Bakken RJ, Winter M. Family characteristics and sexual risk behaviours among black men in the United States
. Perspect Sex Reprod Health
10. Renzullo PO, McNeil JG, Levin LI, et al. Risk factors for prevalent human immunodeficiency virus (HIV
) infection in active duty Army men who initially report no identified risk: a case control study. J Acquir Immune Defic Syndr
11. Levin LI, Peterman TA, Renzullo PO, et al. HIV
-1 seroconversion and risk behaviors among young men in the US Army. The Seroconversion Risk Factor Study Group. Am J Public Health
12. Bray R, Hourani LL, Rae KL, et al. 2002 Department of Defense Survey of Health Related Behaviors Among Military Personnel. Research Triangle Institute Report RTI/7841/006-FR
. Research Triangle Park, NC: Research Triangle Institute; 2003.
13. Bautista CT, Sateren WB, Singer DE, et al. Trends
prevalence among white
and African-American civilian applicants for United States
Military Service, 1985 to 2003. J Acquir Immune Defic Syndr
Keywords:© 2006 Lippincott Williams & Wilkins, Inc.
African American; army personnel; HIV; incidence; Joinpoint; military personnel; trends; white; United States