Article Text

Download PDFPDF

Systematic review and meta-analysis of HIV prevalence among men in militaries in low income and middle income countries
  1. Jennifer Lloyd1,
  2. Erin Papworth2,
  3. Lindsay Grant1,
  4. Chris Beyrer2,
  5. Stefan Baral2
  1. 1Department of International Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, USA
  2. 2Department of Epidemiology, Center for Public Health and Human Rights, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, USA
  1. Correspondence to Jennifer Lloyd, Department of International Health, Johns Hopkins Bloomberg School of Public Health, 615 North Wolfe Street, Baltimore, MD 21205, USA; jwilburn{at}


Objectives To determine whether the current HIV prevalence in militaries of low-income and middle-income countries is higher, the same, or lower than the HIV prevalence in the adult male population of those countries.

Methods HIV prevalence data from low-income and middle-income countries’ military men were systematically reviewed during 2000–2012 from peer reviewed journals, clearing-house databases and the internet. Standardised data abstraction forms were used to collect information on HIV prevalence, military branch and sample size. Random effects meta-analyses were completed with the Mantel-Haenszel method comparing HIV prevalence among military populations with other men in each country.

Results 2214 studies were retrieved, of which 18 studies representing nearly 150 000 military men across 11 countries and 4 regions were included. Military male HIV prevalence across the studies ranged from 0.06% (n=22 666) in India to 13.8% (n=2733) in Tanzania with a pooled prevalence of 1.1% (n=147 591). HIV prevalence in male military populations in sub-Saharan Africa was significantly higher when compared with reproductive age (15–49 years) adult men (OR: 2.8, 95% CI 1.01 to 7.81). HIV prevalence in longer-serving male military populations compared with reproductive age adult men was significantly higher (OR: 2.68, 95% CI 1.65 to 4.35).

Conclusions Our data reveals that across the different settings, the burden of HIV among militaries may be higher or lower than the civilian male populations. In this study, male military populations in sub-Saharan Africa, low-income countries and longer-serving men have significantly higher HIV prevalence. Given the national security implications of the increased burden of HIV, interventions targeting military personnel in these populations should be scaled up where appropriate.

  • HIV

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


National security concerns have been reported for countries with widespread HIV epidemics including the threat of HIV among military personnel affecting troop readiness and incurring substantial costs to military medical care systems. Most countries have mandatory HIV testing for screening new recruits, and while most do offer voluntary HIV testing for longer-serving military personnel, it is not mandatory. However, military personnel are not the only population affected by the virus.

Epidemiological surveillance has demonstrated that female sex workers, their clients, men who have sex with men, and people who inject drugs have specific HIV acquisition and transmission risks, and consequently a higher burden of infection.1–4 While the data on the burden of HIV/AIDS in the militaries around the world are sparse, there is increasing recognition that they are a susceptible population for HIV infection due to their mobility and the tendency to engage in unsafe sexual behaviours. Several studies have been conducted since the late 1990s demonstrating that military physical training programmes decrease soldiers’ immune systems making them more susceptible to infection.5 Militaries also tend to recruit civilians in the 15–25 years age group, an age range when people are at their greatest risk for HIV.6 Countries including Zambia and Namibia have declared that the number one cause of death among their military and police populations is AIDS-related illnesses.7

While on deployment, soldiers are more prone to engage in casual sex, or to pay for sex because they are away from home and from their regular female partners for long periods of time.8 In 2002, a study among the Nigerian Military concerning HIV/AIDS and sexually transmitted infections (STIs) also found that condom use drastically decreased as age increased (from 68.8% at 18–24 years to 37.5% at 45+ years) and the overall frequency of condom usage was only around 50% during peacekeeping operations.9 Another study determined that 32% of the peacekeepers on a mission in Sierra Leone were from countries with a HIV prevalence over 5%.10 Concerns of peacekeepers as a source of incident HIV infections has led to countries refusing help or requesting that the countries should screen their peacekeepers for HIV before deployment.10

Populations within the military also display varying degrees of susceptibility for HIV infection. While women in the military represent a very small portion, they are at increased risk for HIV.11 Women constitute a small proportion of low-income and middle-income countries’ (LMICs) militaries challenging obtaining sufficient data characterising female military data in LMICs. Little is known about the prevalence of same-sex practices among uniformed service members and the HIV prevalence in this population. There are countries around the world that allow openly gay service members to serve in the military including the USA, South Africa, Brazil, Israel and the UK; however, the extent of same-sex practices in these militaries remains unreported in the public domain.12

Many militaries have codes of conduct, policies and regulations, which attempt to restrict sexual practices, including sexual contact with civilian populations while on deployment, same-sex practices and sex with sex workers—and these regulations may drive these behaviours underground, reduce condom access and reduce military personnel's willingness to disclose risky practices to military health staff.13–15 There has been a renewed importance of addressing the needs of people currently living with HIV in order to improve their own health outcomes and minimise the chance of onward HIV transmission, including the use of postexposure prophylaxis and treatment as prevention.16 While militaries in many LMICs provide substantially better healthcare than those available to civilian populations, most are still struggling to provide basic HIV prevention, treatment and care.17

Few countries report the HIV prevalence in their militaries due to concerns for national security; therefore, determining the full extent of the HIV epidemic in LMIC militaries is difficult.18 Over the past two decades, militaries from LMICs have become a key resource for peacekeeping interventions. Countries of militaries that have a high burden of HIV may refuse peacekeepers unless they have been tested for HIV, which could potentiate instability.19 The aim of this systematic review is to determine whether the current HIV prevalence in militaries of LMICs is higher, the same or lower than the HIV prevalence in the adult male population of those countries.


Search strategy and inclusion/exclusion criteria

Peer reviewed and non-peer reviewed articles were included in this systematic review. Peer reviewed articles that were published between 1 January 2000 and 5 December 2012 were collected from PubMed, Embase, Scopus and Ovid. Articles and citations were downloaded from the databases and organised using Endnote X5. Articles were then exported to an Excel file for review during the title and abstract review process. The search terms included MeSH terms (or other associated terms) for ‘military personnel’ cross referenced with terms for HIV and LMIC. The term ‘military’ encompassed a list of different titles including, soldiers, navy, uniformed personnel, army, armed forces, militia and troops.

Other data sources included Clearing-houses, Department of Defense HIV/AIDS Prevention Program, Oxford Journal African Affairs, Africa Today, FHI 360, AIDS Data Hub and Google. These sources were electronically searched using key search terms for HIV and armed forces. The decision was made to include only publicly available articles/reports to ensure the reproducibility of the systematic review.

Studies of any design that had HIV prevalence data for a military were included, even if the study was not primarily about the military. Letters to the editor and conference proceedings were not included. To be included, studies needed to have clear descriptions of the sampling methods and HIV testing methods. Only countries listed as low-income, lower-middle-income or upper-middle-income by the World Bank Atlas Method 2011 were included in the analysis. Studies published in English, French and Spanish were included. All of the articles were subjected to the same criteria at each stage in the analysis. Articles were excluded if there was no prevalence data regarding the burden of HIV among military personnel, the sample size was less than 50 or the sampling methods were unclear.

Screening and data extraction

All of the articles, peer reviewed and non-peer reviewed, were subjected to the same screening process by two independent reviewers (JL and EP) in order to include sources that potentially contained military HIV prevalence data, were not duplicates, and were from low-income, lower-middle-income or middle-income countries. If either reviewer selected an article based on the title, then the abstract was reviewed. The same process occurred for the abstract review. If either reviewer selected an article based on the abstract, the full text was obtained. The full text review was conducted by the same two reviewers. Discrepancies during the full text review were discussed and a consensus was reached as to whether or not to keep the article. After the reviews were complete, data was cleaned and abstracted from the final group of articles using standardised extraction forms that collected data on the average age of the participants, military sample size, total military positive for HIV, accrual methods, sampling methods, country, language, whether or not the study included a biological assessment of HIV, total sample size (including non-military populations), type of military population tested, separate data for men and women, and HIV testing methods/procedures. Methodological quality of the articles was assessed by study sampling and recruitment methods, HIV testing methods and data reporting.

Statistical analysis

HIV prevalence in adults and men aged 15 years and older in each of the countries included in the meta-analysis was obtained using the Joint United Nations Programme on HIV/AIDS (UNAIDS) 2010 Global Report-Annex 1. Data from the US Census Bureau International Division were used as two separate denominators to calculate the HIV prevalence among the military: (1) the number of men aged 15 years and older and (2) the number of men aged 15–49 years.

As a sensitivity analysis, both of the denominators were used because using the data from the 15 years and older age group only may artificially inflate the OR.1 The sensitivity analysis showed that there was no significant difference between the estimates that used these different denominators. Therefore, the denominator of men aged 15–49 years was used for the ensuing meta-analyses since people over 49 years do not contribute a great amount to the HIV infections in LMIC. Female comparisons were not conducted for this meta-analysis since men are still the great majority in militaries around the world.

The meta-analysis compared the odds of HIV among military personnel with the general country populations aged 15–49 years, as well as with the general male populations aged 15–49 years. The Mantel-Haenszel method with random-effects model was used to account for the heterogeneity that occurred because studies from different populations of military members in different countries were used. A correction of 0.5 was added to all zero cells with STATA (V.11). The presence of heterogeneity was tested using the DerSimonian and Laird Q test (τ2 test).

Meta-analyses of subgroups of regions—Africa, Asia, Central America and South America—were conducted to determine if there were statistically significant differences in the military HIV prevalence by region. Meta-analyses of subgroups of countries by income level were also conducted. The following World Bank Atlas Method categorisation scheme was used to classify countries; five low-income countries ($1025 or less gross national product (GNP)), four lower-middle-income countries ($1026 to $4035 GNP), and two upper-middle-income countries ($4036 to $12475 GNP).20

A sensitivity analysis compared the OR of a meta-analysis that included new recruits and people who had been in the military for longer than 1 year to a meta-analysis that only included studies with people who had been in service for longer than 1 year. This was conducted to determine if the new recruits affected the results of the meta-analysis due to factors associated with their age (ie, less sexual activity, healthier, etc). As of 2010, Afghanistan, Timor-Leste and Ethiopia were not included in the UNAIDS Global HIV report. Therefore, studies from these countries were excluded from the systematic review.


The search criteria identified 2064 peer reviewed articles and 150 non-peer reviewed articles for the title review (figure 1). The final group of 13 articles and five surveillance reports represent 147 591 armed forces members including police, defence forces, army, new military recruits, air force, navy and customs (some differentiated, some not) in 11 countries: one from Central America, five from Asia, one from South America and four from sub-Saharan Africa (SSA). Out of the 11 countries included, 5 were low-income countries (Burma, Cambodia, Guinea, Guinea-Bissau and Tanzania), 4 were lower-middle-income countries (Belize, India, Nigeria and Vietnam) and 2 were upper-middle-income countries (Brazil and Thailand).

Figure 1

Search strategy flow chart.

Table 1 provides summary statistics for each country, including HIV prevalence for the military, general population and male population, military sample size and ORs. The overall military HIV prevalence was 1.08% (95% CI 0.58 to 1.59), with the country-specific military HIV prevalence ranging from 0.06% for India (95% CI −1.24 to 1.36) to 13.8% in Tanzania (95% CI 10.35 to 17.31) showing significant variation by region. Table 2 provides a summary of the data by region and provides a subgroup analysis of the data by length of service and country income level. Central America and South America were represented by one country each thereby preventing any regional meta-analyses. A subgroup analysis was not conducted for upper-middle-income countries because there were only two countries in that group.

Table 1

Meta-analyses of aggregate country data comparing HIV prevalence among military members and general and male populations in low-income and middle-income countries, 2000–2012

Table 2

Subgroup meta-analysis of OR for HIV prevalence among military for region, income group and length of service

The overall estimate for the OR for a military member to be living with HIV as compared with the general population in LMICs is 1.14 (95% CI 0.50 to 2.57). None of the general country population ORs were significant when subgroup analyses were conducted at the region level. The overall estimate for the OR for a military member to be living with HIV as compared with the male population in LMICs is 1.21 (95% CI 0.54 to 2.74). The ORs comparing military populations with the general population and the male population in low-income countries were statistically significant OR: 3.39 (95% CI 2.62 to 4.40); OR: 3.96 (95% CI 2.99 to 5.25), respectively (table 2).


This systematic review of HIV among uniformed service members evaluated the current burden of HIV with a comparison with the general population across country-level economic status. Our data reveals that within the different settings, military HIV burden may be higher or lower than the civilian male populations. Overall, the pooled OR comparing HIV among military populations with reproductive age adults (OR: 1.14, 95% CI 0.50 to 2.57) and with that of other men (OR: 1.21, 95% CI 0.54 to 2.74) were not significantly elevated in the countries included in the analysis (table 1).

There was geographical variation of the results with a significantly elevated burden of HIV in military personnel across the four SSA countries included in this systematic review (OR: 2.81, 95% CI 1.01 to 7.81) when compared with reproductive age men. Only four studies meeting inclusion and exclusion criteria were available from SSA limiting generalisability across the continent. However, the data presented here are consistent with an earlier study with data collected until 2006 that focused on characterising the burden of HIV in African militaries.40 These data do suggest that the epidemic setting in which men are serving may be determinative of the risks of HIV acquisition. Moreover, the meta-analyses by subgroup of income level found that the OR for HIV in low-income country militaries compared with reproductive age men was 3.96 (95% CI 2.99 to 5.25). Since three of the five low-income countries were from SSA, this is likely a function of the HIV prevalence in SSA but underscores the need for expanded resources in these settings.

The results of this systematic review show that longer-serving military members in LMICs, men in SSA militaries and militaries serving in low-income countries have increased odds of HIV compared with the general country population. Local, national and regional deployments are opportunities for population mixing and for structural risks, including decreased condom access and antiretroviral therapy (ART) interruptions. The burden associated with HIV-related morbidity and mortality creates a need for additional resources to train new recruits.10 Countries that refuse to publish or share data on HIV in their militaries disrupt and prevent HIV programme evaluation for the soldiers, as well as the civilians with whom they interact.19

Our results show that military populations are not universally at higher risk for HIV acquisition. The lack of uniformity across countries with respect to military HIV prevalence may be related to multiple factors. If men in middle-income countries are not serving as often in low-income settings with heightened background HIV prevalence, then their risk of HIV acquisition from sex in these settings is limited. Instead, their primary source for HIV acquisition would be from same-sex practices with men in their own military. In this study, new recruits had lower HIV prevalence, which may be a function of age; however, the results suggest that acquisition is occurring during service. Changing sexual behaviour is difficult to achieve, especially for off-duty soldiers and deployed peacekeepers.6 Transmission of STIs can be prevalent during and after conflicts, and HIV is up to 20 times more likely to be transmitted when a person has a STI.6 Peacekeepers living with HIV from high HIV prevalence countries have the potential to spread HIV to the community, and peacekeepers sent to countries with high HIV prevalence may be at risk of acquiring HIV infection.41 Unfortunately, most HIV programmes in militaries in LMICs are limited in scope and reach.6 Changing behaviour is difficult so prevention during deployment (ie, providing condoms, and adequate access to safe and confidential voluntary counselling and testing (VCT)) will aid in decreasing HIV transmission.

In the era of effective ART, men living with HIV may be ready, willing and able to serve in military contexts, but also require military healthcare systems to be responsive to their needs. Appropriate and targeted prevention programmes, including voluntary medical male circumcision for soldiers in countries where circumcision is part of national strategies, such as Tanzania, should be considered. As HIV responses have changed in relation to new intervention models such as treatment as prevention, and efforts to decrease stigma and expand testing and care for men (who are underserved compared with women across SSA), HIV programmes in militaries of high burdened countries are well positioned to adapt HIV prevention and control approaches within their structures.42 ,43 HIV programmes need to address the health needs of military men in countries such as Thailand, Brazil and many high-income countries (the USA, the UK, Australia) where gay and bisexual men may now serve openly. This will be challenging for many militaries, especially where these men and their behaviours are criminalised, but nevertheless, this will be an important strategy to preserve military health in the future.

The sensitivity analysis comparing the studies that only included longer-serving military members had a statistically significant higher OR compared with the general population. The results of this sensitivity analysis hint at the possibility that the military environment does increase the risk of HIV relative to the length of time a man has been in service. Previous reports have found this relationship and ascribed it to frequent deployments, casual sex and other high-risk practices over time.8 As the length of service increases, the number of times these activities potentially occur also rises, leading to increased odds of HIV acquisition.

This study had several limitations. Publication bias is a relevant limitation since many countries could have conducted military HIV prevalence studies but not published the results for a variety of reasons, including national security concerns. This appears to be the case for Botswana with approximately 9000 active duty personnel—a country with a broadly generalised HIV epidemic among reproductive age men.4 Article accessibility is another limitation. To be included in this systematic review, articles needed to be in the public domain. Fifteen non-peer reviewed articles were selected for data abstraction; however, nine of those were excluded because the full article could not be accessed.

Other limitations include the range of years included in the systematic review and the pooled analysis. The year 2000 was used instead of 2005 because there were not enough studies conducted between 2005 and 2012 for stability in the estimates. Since 2003 marks the beginning of ART availability in much of Africa, studies in the 2000–2003 period may have lower HIV prevalence due to higher morbidity and mortality among men. The data was pooled across different regions with different HIV epidemic dynamics for the sensitivity analyses resulting in unmeasured confounding, limiting causality assessments.

It is a reality of our time that military conflicts continue to occur and that regional and international forces are now increasingly deployed in response. HIV among military personnel is now clearly a transnational issue, but one for which the future is hopeful. HIV is treatable, and military medicine has a long and proud history of high quality services in many countries, suggesting that the continuum of HIV care, from prevention, regular testing and reliable ART are achievable goals for military health systems.

Key messages

  • There has been no recent systematic review conducted on HIV prevalence in militaries in low-income and middle-income countries.

  • In this study, only male military populations in low-income, sub-Saharan African countries and longer-serving men have significantly higher HIV prevalence.

  • In order to preserve relations and decrease tensions between and within countries at times of peacekeeping missions, HIV in militaries needs to be addressed.


The Johns Hopkins Center for AIDS Research (CFAR) provided salary support to CB during the development of this work. In addition, the Foundation for AIDS Research (amfAR) provided support for the searches completed here.



  • Handling editor Jackie A Cassell

  • Contributors SB, EP and JL developed the search protocol which was implemented by JL. JL and EP reviewed articles and abstracted data. JL and EP read the articles. JL drafted the manuscript. LG, CB, SB and EP edited and revised the manuscript.

  • Funding The Johns Hopkins Center for AIDS Research (CFAR) provided salary support to CB during the development of this work. In addition, the Foundation for AIDS Research (amfAR) provided support for the searches completed here. The funders did not have any input to the design or implementation of the study or the decision to publish the findings.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.