Prevalence of Zoonotic and Vector-Borne Infections Among Afghan National Army Recruits in Afghanistan

Introduction

Acute febrile illness causes measurable morbidity and mortality in Afghanistan, potentially due to a variety of described vector-borne and zoonotic infections, including Plasmodium vivax (PV) and Plasmodium falciparum (PF) malaria, visceral and cutaneous leishmaniasis, and febrile illness secondary to bunyavirus, flavivirus, and Brucella spp. infections (Wallace et al. 2002, Aronson et al. 2006, Mofleh and Ahmad 2012, Elyan et al. 2014, Leslie et al. 2014). The majority of cases are detected through individuals with acute febrile illness accessing care and receiving diagnostic testing, much of which is not available outside regional and national reference laboratories or outside focused studies (Elyan et al. 2014, Leslie et al. 2014). Access to appropriate diagnosis and treatment is further compromised by frequent over-diagnosis of malaria, a commonly attributed cause for febrile illness (Leslie et al. 2012, Reynolds et al. 2013). Other etiologies for febrile illness may be under-recognized secondary to presumed arthropod vector geographic distribution. In 2008, an outbreak of Crimean-Congo hemorrhagic fever (CCHF) was described in Afghanistan, but it is not a frequent consideration in acute febrile illness differential diagnosis in the country (Mustafa et al. 2011, Mofleh and Ahmad 2012). Brucellosis, a zoonotic infection endemic to the Middle East and Central Asia, is reputed to be common in Afghanistan, but there are no recent data regarding population prevalence for illness or serologic evidence of exposure (Wallace et al. 2002). Thus, accurate data for zoonotic and vector-borne infection cases are limited due to lack of specific diagnostic testing at primary care centers, subacute or asymptomatic infections not presenting for care, and diagnosis and care availability only at facilities that are far from case acquisition.

One approach to improving differential diagnosis is by improving knowledge of the geographic distribution of these infections. In the 2008 CCHF outbreak, 11.2% of asymptomatic individuals who were household contacts of cases had CCHF virus antibody (Mustafa et al. 2011). There are few data regarding asymptomatic or general population prevalence for these infections; a study from Iran detected 1.6% prevalence of PV in blood smears taken from 446 asymptomatic Afghan immigrants in Iranshahr district, whereas a seroprevalence assessment of deployed British troops in Helmand revealed no samples reactive for CCHF antibody (Nateghpour et al. 2011, Newman et al. 2014).

Opportunities for serologic surveillance in the general population are limited, though banked serum from nationally representative groups provides the ability to estimate general population prevalence and the geographic distribution of pathogen exposure. One such population group are the armed forces, who may provide samples for routine testing during health checks (Brown et al. 2011, Ribeiro et al. 2012, Wickremasinghe et al. 2014). In 2010–2011, a large cross-sectional seroprevalence assessment of 4750 randomly selected Afghan National Army (ANA) recruits was conducted to assess prevalence and correlates of HIV, syphilis, herpes simplex-2 virus, and hepatitis B and C virus (Todd et al. 2012). Serum samples from participants consenting to additional testing from de-identified specimens were stored and available for public health testing. The Disease Early Warning System department within the Afghan Public Health Institute of the Ministry of Public Health prioritized testing for CCHF, visceral leishmaniasis (VL), dengue virus (DENV), PV and PF malaria, and brucellosis from a representative group of specimens. The purpose of this analysis is to measure the prevalence of markers of past exposure to or current infection with the selected vector-borne and zoonotic pathogens and to describe the distribution of detected cases among a nationally representative sample from male ANA recruits.

Materials and Methods

The ANA seroprevalence study methods have been previously described (Todd et al. 2012). Briefly, randomly selected male recruits presenting at the Kabul Military Training Center for medical screening and enlistment were approached for study participation. Eligible participants were male, aged 18 years or older, Dari or Pashto speaking, and able to provide informed consent. Consenting participants completed an interviewer-administered questionnaire detailing sociodemographic characteristics and risk behaviors, provided an intravenous specimen for testing and serum storage, and received pre- and post-test counseling. Participant names were not recorded; assigned study numbers were the sole identifier for all study materials. Approval for the original study and the amendment for this nested analysis were obtained from the institutional review boards of the Columbia University, the Walter Reed Army Institute of Research, and the Ministry of Public Health of the Islamic Republic of Afghanistan before performing testing.

For this nested sub-study, study numbers of participants consenting to use of their donated specimen for additional public health testing were combined into a separate database. The database was stratified by province of birth or having been born outside Afghanistan, and 25 study numbers selected randomly from each group. Random selection was performed by taking the total number of participants from each province and then selecting 25 random numbers from that total (www.randomizer.org). Five additional numbers were selected to hold in reserve to replace initially selected samples with insufficient serum for testing. Four provinces (Zabul, Nimroz, Uruzgan, and Nuristan) had fewer than 25 participants; all samples from these provinces were tested in this analysis.

Results

The demographic characteristics and travel history of participants contributing 809 samples for this analysis are displayed in Table 1. Generally, about two-thirds of participants had previously lived outside Afghanistan, usually in either or both Pakistan and Iran. In the screening tests, the overall prevalence of antibody to DENV (2.8%, n = 23), CCHF (4.1%, n = 33), and VL (1.0%, n = 8) was low. For malaria species, PV and PF antigens were detected in 2.1% (n = 17) and 4.3% (n = 35) of participants, with 0.5% (n = 4) reactive for both. For Brucella spp., 8.0% (n = 65) had titers >1:40 by slide agglutination testing. Distribution of reactive pathogen tests by region of birth is displayed in Figure 1.

OPEN IN VIEWER

FIG. 1.

Case distribution of select zoonoses/brucellosis by region of Afghanistan among Afghan National Army recruits (n = 809). CCHF, Crimean-Congo hemorrhagic fever; PRNT, plaque reduction neutralization test; JEV, Japanese encephalitis virus.

For DENV, 60 specimens received PRNT testing, including the 23 with reactive results by ELISA and a further 37 randomly selected samples from this nested study population with nonreactive ELISA results for validation purposes. Specific DENV subtype distribution by PRNT assay for all 60 specimens is displayed in Table 2. Among specimens nonreactive by ELISA, all those with evidence of DENV by PRNT were positive for subtype 3. For malaria, only 7 out of 56 reactive samples had detectable parasitic DNA. Specifically, one sample had DNA detected for both PV and PF though it was reactive only for PF antigen by RDT, whereas 17% (n = 6) of samples with RDT reactive for PV had detectable DNA.

For Brucella spp., 83.1% (n = 54) of samples with titers >1:40 by slide agglutination also had titers >1:80 by tube agglutination. Due to lack of statistical power from the small sample size, only confirmed brucellosis infection was assessed for correlates. Those born in Kabul and the surrounding provinces had lower odds (OR = 0.19, 95% CI: 0.04–1.00) of having Brucella spp. antibody compared with other provinces, with those born outside Afghanistan serving as the reference. No socioeconomic indicators were associated with Brucella spp. antibody (data not shown).

Discussion

Key findings from this analysis include a nearly national distribution and a relatively high prevalence of Brucella spp. antibody, a prevalence >10% of CCHF antibody predominantly in the northern and western regions, and evidence of flavivirus exposure with similar prevalence between regions in Afghanistan and those having been born outside Afghanistan. This effort represents the first attempt to describe national distribution of priority zoonotic and vector-borne pathogens; however, certain limitations should be considered when interpreting these data. Military populations, due to their national representation, are ideal for estimating exposure risk and prevalence, with important caveats. The ANA recruit population was nominally healthy, though a review of systems to elicit recent fevers and other symptoms of the specified infections was not included in parent study data or in routine medical screening conducted by the ANA at enrollment. Further, no questions regarding recent insect stings/bites or exposure to or participation in slaughter or processing of animal carcasses were asked during routine screening or by the parent study. Participant vital signs were also not measured and it is possible that some recruits may have been febrile at the time of study entry, though they otherwise appeared well.

Because most laboratory results represent evidence of only prior exposure, there is no way to ascertain the geographic location at which the individual was exposed. Among the tested population, 62% reported previously living outside Afghanistan and some infections may have been acquired during that time. Further, though test result regions are determined by birth province, it is possible that some individuals did not recently reside in or repatriate to their native provinces. Infection distribution data are provided by region, as determined by reported province of birth, to reduce this effect. Last, several of the screening tests would have detected active infection/recent exposure only, particularly the histidine-rich protein 2 (HRP2) antigen for PF malaria, which remains present for up to 3 weeks after resolution of infection. Additionally, for pathogens for which screening and confirmatory testing was available, screening tests may have over-estimated prevalence, particularly for malaria and DENV. This may also have been the case for pathogens where only a single test was used. For example, CCHF ELISA results should be interpreted with caution, as no confirmatory assay was performed and cross-reactivity with antigenically similar nairoviruses is possible. Further work may be justified to clarify whether other nairoviruses are circulating in Afghanistan.

With those caveats, some interesting patterns emerge with regard to regional distribution of pathogen exposure. The results suggest that exposure rates to key causes of acute febrile illness are low, yet measurable, and not evenly distributed through the country. For example, our data describe a higher percentage of reactive CCHF tests among those originating in the north and west; whereas brucellosis is fairly evenly distributed throughout the country, and mosquito and sandfly vector-borne infections are generally more prevalent in the southern region (Leslie et al. 2006, Adimi et al. 2010).

An antibody to Brucella spp. detected by slide agglutination, indicative of prior exposure, was present in all regions and among those born outside Afghanistan (largely in Pakistan and Iran), consistent with its endemic distribution across Central Asia (Pappas et al. 2006, Dürr et al. 2013). Confirmed flavivirus antibodies were largely clustered in southern and eastern regions comprising provinces with populations with a high mobility to and from Pakistan, or among those born outside Afghanistan, consistent with endemicity of DENV and JEV in Pakistan, where many participants were born or had previously resided (Erlanger et al. 2009, Hasan et al. 2014). A recent flavivirus prevalence study from Afghanistan identified an antibody for West Nile virus, JEV, and DENV among symptomatic patients from several southern provinces bordering Pakistan and from Kabul (Elyan et al. 2014). Antibody detection for JEV has been documented in Helmand (south) and Kunduz (northeast) provinces, with DENV additionally identified in Helmand three decades earlier (Wallace et al. 2002).

A relatively high level of reactivity to the VL K39 epitope was observed in the northeast region, consistent with case distribution from a 2006 report (Leslie et al. 2006). This was particularly true for samples among those hailing from Nuristan, a remote eastern province with very limited access to healthcare due to poor transportation infrastructure and chronic insecurity; there are sparse health data regarding clinical leishmaniasis cases for this province. Further assessment is needed to determine whether VL represents a serious health threat in this region.

Conclusions

This analysis highlights the need for geographic and temporal correlation with case reporting and screening, as the results from this study cannot be linked to the site of exposure. The findings also lead to a call for further monitoring of vaccine preventable infections such as JEV. Additionally, continued surveillance for other vector-borne etiologies will be necessary to monitor the effectiveness of control measures, such as the distribution of long-lasting insecticide-treated bednets, indoor residual spraying, and the use of repellants. Finally, enhanced point-of-care (POC) diagnostics are essential to identify diseases such as brucellosis, leishmaniasis, and malaria to allow for effective treatment and prevention of long-term sequelae. Greater availability and accuracy of POC diagnostics deserves particular emphasis, as the settings in which cases arise are often remote where lack of capacity to perform ELISAs and molecular diagnostics is not available and cost and limited infrastructure (lack of electricity and refrigeration) remain as barriers to introducing these technologies. These data indicate that further improvements are needed in the diagnostic accuracy of commercially available assays for the assessed pathogens; more than 25% of those nonreactive for DENV IgG by ELISA had evidence of flavivirus antibody by PRNT assay. Since these diseases infect both humans and animals and cross-transmit, inter-sectoral collaboration is recommended to ensure a One Health Approach in which the animal health, human health, and environment protection sectors collaborate to address zoonotic diseases. Future studies need to focus on both animals and humans to assess potential reservoirs of infection and to determine prevalence of and risk factors for transmission of zoonotic diseases in this region to provide an evidence base for public health policies and infectious disease control programs.