Investigation of West Nile Fever Seroprevalence and Associated Risk Factors among Horses in Egypt

Determination of the seroprevalence and risk factors associated with West Nile Fever (WNF) in horses is essential for adoption of effective prevention strategies. Our objective was therefore to determine the seroprevalence and identify the risk factors associated with West Nile virus (WNV) infection in the most densely horse populated provinces in Egypt. A cross-sectional study was conducted on 930 horses distributed over ve provinces in the Nile Delta of Egypt in 2018. The randomly selected horses from geographical studied areas were serologically tested for WNF, and the horses’ information was obtained from owners. Four variables (geographic location, breed, gender, and age) were used for risk analysis. A total of 156 (16.8%) serum samples were serologically positive for WNV. The highest prevalence of WNV infection was between mixed breed (21.5%), male (26.4%) horses with age ≥ 15 years (68.1%). Age showed the highest risk for the infection with WNV (OR = 4.3, 95% CI 3.0 to 6.2). Gender and breed also showed important risk factors for WNF infection in horses (OR = 2.4, 95% CI 1.6 to 3.7; OR = 1.9, 95% CI 1.2 to 2.8, respectively). However, geographic location seems to have no impact on the seroprevalence of WNV infection among horses in the Nile Delta of Egypt. Our ndings strongly recommend intensive surveillance and implementation of effective control and prevention strategies against WNF especially in male of mixed breed horses with ages ≥ 15 years.


Introduction
West Nile Fever (WNF) is a zoonotic, widespread vector-borne infection caused by West Nile virus (WNV) that belongs to the family Flaviviridae 1 . The disease is listed by the World Organization for Animal Health (OIE) as noti able disease that requiring the individuals and the member countries to report its occurrence. West Nile Virus maintained in the nature and spread across the country by cycling between as a biological vector and birds, the primary host, taking advantage of free-ranging birds and mosquitoes, and incidentally infects humans and horses that do not contribute to the spread or ampli cation of the virus 2,3 . In Egypt, Culex antennatus is the major mosquito species responsible for transmission in the ampli cation cycle 4 . All over the world, more than 300 bird species may act as vertebrate hosts for WNV, and the infected migratory birds are thought to play the major role in the spread of the virus to the free areas 5 .
The WNV was rst reported in 1937 in the West Nile region of Uganda and over the past decades, the virus became endemic in many African including the Middle East, European, and Asian countries 6 . Since the 1960's, Europe has reported sporadic cases and outbreaks of WNV in both human and horses.
Surveys in parts of Europe and the Middle East have shown that up to a third of the tested horses have been exposed to the virus even with absence of the clinical disease 7,8 . In 1999, the WNV was rst reported in the eastern United States where it caused severe epidemic of meningoencephalitis in New York City. Then, it spread throughout Canada, and Central America 5 . In the early 1950s, WNV was rst recorded in the Mediterranean basin, including Egypt and Israel, and equine WNV encephalomyelitis was rst recorded in Egypt in 1959 4,9 . Several outbreaks among equine populations was then sparked the importance of the disease and raise interest in predicting where and when the virus would appear next 4,10 . However, few small-scale studies conducted to examine the prevalence of WNV infection among equine in Egypt.
De ning the risk factors of acquiring infection with WNV is a key component of successful prevention and control strategies and help in the decision of whether or not to vaccinate 11 . Intriguingly, animal related factors such as age and breed showed association with the seroprevalence of WNV infection among horses 12 . However, the effect of these factors is still controversy and need further investigation.
The main objective of this study was therefore to investigate the seroprevalence of WNV infection among horses raised in the Nile Delta of Egypt, as well as to determine the risk factors associate with the infected horses.

Seroprevalence of WNV
The seroprevalence of BLV was determined in 930 serum samples obtained from three different breeds of horse (Arabian, Thoroughbred, and Mixed) located in 5 provinces in the Nile Delta of Egypt (KF, GB, MF, Qal, and GZ) at age ranged from 2 to 16 years. The distribution of horses based on the risk factors was illustrated in Table 1.  Table 2 and Fig. 1. The distribution of WNV-positive horses was differed between age (P < 0.001), gender (P < 0.001), and breed (P < 0.01; Table 2). Age (year) < 2. The results of this study showed moderate associations between the seroprevalence of WNV infection and age (Phi Coe cient and Cramer's V = 0.56), and weak association with gender (0.28). However, the Phi Coe cient and Cramer's V of 0.10 and 0.06 indicated no association between the WNF seroprevalence and bread and geographic location, respectively.
The stepwise logistic regression model indicated that age, gender, and breed were signi cant risk factors for WNV infected horse (Table 3). Geriatric horses above 15 years had greater number odds for seropositivity of WNF (OR = 4.3, 95% CI 3.0 to 6.2; Fig. 2), where the horses with age of 15 years increased risk (90%) of being seropositive for WNF and interestingly the probability of risk increases with age ( Fig. 3). Male horses and mixed breed are also at a high risk for WNV infection (OR = 2.4, 95% CI 1.6 to 3.7; OR = 1.9, 95% CI 1.2 to 2.8, respectively; Fig. 2,3). However, geographic location was not associated with increased risk of being seropositive for WNV (OR = 0.9, 95% CI = 0.8-1.1; Fig. 2,3).  Fig. 1. These provinces were selected as they are considered the most densely horse populated areas in Egypt, where the horses are mainly raising for horse-drawn carriages transport and horse-based tourism. The climate of the selected areas is dry summers and mild, wet winters that help in multiplication and distribution of the vectors.

Animals and sample collection
Adequate sample size was used for this study based on calculation using Cochran's formula 13 , as follow: where n is the sample size, Z is the statistic corresponding to level of con dence, p is expected prevalence, and e is precision (corresponding to effect size). The level of con dence was used is 95%, and the expected prevalence used in this study was 15% based on the reported prevalence published in Zaher and Ahmed (2014). The precision (e) used in this study was 5% based on 14 .
A total of 930 animals were screened for WNV during 2018. Blood samples were collected from 500 males, 300 females and 130 Gelding with age groups ranged from < 5 to > 10 years. The blood samples were collected from jugular vein using 20G needles and 10 mL blood collection tubes. The proposed puncture site was cleared of debris and swabbed with gauze containing 70% isopropyl alcohol. Serum was collected after centrifugation of the blood samples at 1500 rpm/min for 10 min. The serum samples were stored at -20 C for serological testing.
3.3. Serological testing using ELISA All serum samples were serologically testing for WNV using ID screen West Nile Competition ELISA kit (IDVet, Montpellier, France) according to the manufacturer`s instructions.

Statistical analysis
The associations of WMF with different risk factors were evaluated using the Cochran-Armitage trend test using PROC FREQ of SAS 15 . The strength of associations was evaluated through Phi and Cramer's V value. Chi-square and stepwise forward multivariable logistic regression were used to identify the most important risk factor(s) associated with WNF. The P-values for entry into or removal from the logistic regression models were < 0.05. The logistic model, tted with WNF as the outcome variable (present: 1, absent: 0), included xed effects of the risk factors of breed (3 levels: Arabian, Thoroughbred, and Mixed), gender (3 levels: male, female, and gelding), age (5 levels: <2.5, 2.5-<5, 5-<10, 10-<15, ≥ 15 years), and geographic location (5 levels: KF, GB, MF, Qal, and GZ). The stepwise elimination process was stopped once all remaining variables were signi cantly (P < 0.05) contributing to the model. The t of the multivariable logistic regression model was assessed using the Hosmer-Lemeshow goodness-of-t test. A logistic regression model predicts the log odds (logit) for outcome as an additive function of the risk factors. The odds ratio was used as an approximate measure of relative risk (the likelihood of having WNF in animal with a given risk factor compared with animal without the risk factor). The 95% Con dence intervals (CI) for odds ratio estimates were obtained as described by Lemeshow and Hosmer (1984). Odds ratios greater than 1 indicate an increased risk of the outcome (WNF) with increasing value of risk factor and odds ratios less than 1 indicate a decreased risk of the outcome (WNF) with increasing value of risk factor (Lemeshow and Hosmer 1984). Statistical analyses were performed using SAS 9.4 (SAS Inc., Cary, NC).

Discussion
West Nile virus is a mosquito-borne pathogen of global public health importance. With the rapid global spread of WNV and the endemic state, WNV might spread in new geographical areas in Egypt. To the best of our knowledge, this is the rst large-scale study investigated the WNV seroprevalence between horses and the associated risk factors in Egypt. Therefore, the results of this study provide updated information about the WNV seroprevalence and the associated risk factors that can be used as a guide for effective prevention strategies. The major advantages of the current study are: 1) broad context of many provinces, placing it amongst the few studies that have examined the seroprevalence of WNV infections in horses across the most densely equine populated areas in Egypt; and 2) large number of participating horses that re ects on the robustness of the association between variables. The main nding of this study is that the male mixed breed horses with ages ≥ 15 y are more susceptible to WNV infection in Egypt.
West Nile virus circulating in Egypt since 1951s 9 . In Egypt, most of the earlier studies were focused on the investigation of WNV epidemiology in humans 9,16,17 16 . The higher temperature above 30 °C as in the Nile Delta of Egypt particularly in summer, and low rainfall during winter and spring were associated with reduction in the mosquito activity and decrease in its growth. Consequently, the spread of WNV among horses will be decreased because the propagation of WNV in mosquitoes is temperature dependent 27  The results of this study didn't show difference in the seroprevalence of WNV infection between studied geographic regions in the Nile Delta of Egypt in agreement with earlier studies 16 . This is a sensible result because the similar geographic nature of the land in the ve studied provinces that is characterized by at low-lying areas. Consequently, the insect population density, the main transmission vector for BLV, are approximately similar in these provinces. Another possible reason for the non-signi cant difference in the seroprevalence of WNV infection among the studied geographical areas is that the equine trading between provinces is not common in Egypt.
The results of the current study showed marked increase in the risk of WNV infection with age suggests recurrent circulation of WNV. This result is consistent with earlier studies 12,32,33 . However, other studies didn't show any association between the seropositivity of WNV infection and age 10,18,21 . It has been reported that the older affected horses have been more likely to die from the WNV infection [34][35][36] . In humans, the median age of affected human is 63 y, suggesting the older people are more prone to acquire WNV infection and approximately 20-fold increased risk at age > 60 y for susceptibility to neurological involvement and retinopathy associated with severe disease 32,33 . The reason behind this positive relationship between seropositivity of WNV infection and age is unknown but the older horses > 10 years are might be immunocompromised animals, where aging is associated with a dysregulation of the immune system 37 . With age, alterations occur to adaptive immunity compounds and reduction in innate recognition and signaling resulting in increased susceptibility to WNV in elderly horse 38,39 .
The second risk factor for WNV infection reported in this study was the gender. Our ndings come in accordance with Epp, et al. 34 that found a higher seroprevalence of WNV infection among stallions than mare or gelding. This interesting nding has been reported in human where they reported a higher seroprevalence of WNV in male than female 40 . Additionally, a higher rate of hospitalization and a higher incidence of neuroinvasive disease was shown in WNV-infected males than WNV-infected females 41 .
The reason behind this result is unclear but several human studies reported gender deference in the seroprevalence of WNV 40,42 . They suggest that gender may play an important role in sustaining WNV immunity. A marked difference was found in immune response against WNV infection between males and females as the infection progressed, with WNV-infected males exhibiting an altered cytokine response 42 . The difference in immune response to viral infection between males and females may be partly attributed to the sex hormones estrogen and androgen 43 . It has been reported that stallions have poor immune response in comparison with mare or gelding because the testosterone hormone impairs immune system functioning 44 . Furthermore, due to culture believe, the male horses are more preferable for pulling large wagon and horse-drawn carriage that make them more vulnerable to mosquitoes and WNV infection.
Intriguingly, the mixed breed horses showed higher seroprevalence of WNV than the Arabian and Thoroughbred horses in agreement with other earlier study 21 . However, different study reported no effect of breed on the seroprevalence of WNV infection in horses 18 . The reasons for the higher WNV seroprevalence found in our study among mixed breed horses, as compared to other breeds, is unknown but it might be due to different preference of mosquitoes to different breeds of horses, due to variables such as density of hair and differences in sweat composition 20 .
The limitation of the present study warrant mention is that this study is a cross-sectional study that is require a larger sample size and more geographic representation than was the case in this study and therefore to verify the associations obtained in this study, longitudinal studies are required.

Conclusions
The results of the present study con rm the circulation of WNV among horses in Egypt and provide updated information concerning the geographic area affected by the virus. Our ndings strongly recommend intensive surveillance and implementation of effective control and prevention for male of mixed breed horses with ages ≥15 years because they are at a high risk for WNV infection.

Declarations
Con icts of Interest territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.