Peste des petits ruminants (PPR) in Africa and Asia: A systematic review and meta‐analysis of the prevalence in sheep and goats between 1969 and 2018

Abstract Background Peste des petits ruminants (PPR) is a prevalent viral disease of sheep and goats that impacts productivity and international animal trade. Despite the substantial economic consequences related to PPR, little is known about the prevalence of this disease at the broad geographical levels. Objective The present study aimed to use a systematic approach to assess the regional prevalence of PPR in sheep and goats, and the associated factors that contribute to prevalence estimates. Methods Published articles on PPR in sheep and goats were searched in PubMed, Web of Science, Scopus, Google Scholar and the reference lists of articles reporting the prevalence from 1 January 1969 to 31 December 2018. Articles were selected using inclusion and exclusion criteria. Since the heterogeneity among the studies was significant, pooled prevalences were estimated by a random effect meta‐analysis model. Results Data on the prevalence of PPR were obtained from Africa and Asia, where the pooled prevalence estimates were 40.99% (95% CI: 37.20%–44.79%) and 38.43% (95% CI: 35.64%–41.22%) respectively. Overall, the estimated pooled prevalence at Africa‐Asia level in sheep was 39.31% (95% CI: 35.75%–42.88%) and in goats was 39.57% (95% CI: 36.66%–42.48%). Significant heterogeneity (I 2 > 80%) was noted in most pooled estimates. Conclusion The results on the regional prevalence estimates of PPR presented here will be useful in raising awareness and advocating for Governments to engage in initiatives to eradicate PPR and prevent it from spreading to other continents.


| INTRODUC TI ON
Peste des petits ruminants (PPR) is an economically significant and widespread viral disease of ruminants that is caused by Peste des petits ruminants virus, a Morbillivirus that belongs to the family Paramyxoviridae (Gibbs, Taylor, Lawman, & Bryant, 1979). PPR spreads quickly in susceptible ruminant species, and the highest number of outbreaks occurs in sheep and goats. Cattle, camels and several wild ruminants have been infected occasionally; however, there is currently no evidence to show that the disease is maintained in these populations without concurrent infection in sheep or goats (Lembo et al., 2013).
PPR is considered to be the most significant economic threat to the development of sustainable sheep and goat production across the developing world, particularly in Africa and Asia. For example, in India, the estimated annual loss caused by PPR in sheep and goats was about US$ 1,297 million per year (Singh, Bardhan, Verma, Prasad, & Sinha, 2014), while in the Turkana County of Kenya, it was US$ 19.1 million . Moreover PPR affects national and international movement and trade of sheep and goats and their products. The disease is currently considered as one of the main transboundary and notifiable disease that constitutes an emerging or re-emerging threat in many countries of the world. In March 2015, PPR was targeted as a high priority disease for progressive control by the World Organisation for Animal Health (OIE) and the Food and Agriculture Organisation (FAO) to eradicate the disease by 2030.
PPR is also known as sheep and goat plague. The causative agent, Peste des petits ruminants virus (PPRV), is considered sensitive to abiotic environmental factors, and it does not survive long outside of a host. The virus is primarily transmit through aerosol and direct contact between infected and susceptible animals (Fournié et al., 2018).
The incubation period of the disease is 4-6 days, but can be up to 14 days. Clinical infection varies, and may include fever, oculo-nasal discharges, oral erosions, pneumonia and diarrhoea (Naznin, Ahaduzzaman, Chowdhury, & Biswas, 2014). The infection period is usually 5-7 days, and death of the infected animal may occur within 10-12 days post-infection due to severe dehydration and respiratory failure (Diallo et al., 2007). Morbidity and mortality are usually high, and PPR can create epidemics that can cause up to 100% mortality in susceptible sheep and goat populations (Parida et al., 2016).
In some cases, particularly in the mild form of the disease, affected animals develop coughing and diarrhoea, and spontaneous recovery may occur within 10-15 days of infection. The magnitude of the disease depends on several factors such as the virulence of the PPR virus strain, species of animal, age, gender, breed, host immune status and previous population exposure to PPRV (Abubakar, Rasool, et al., 2016). Sheep and goats in endemic regions may develop lifelong immunity following natural infection, but naive animals may allow continuous circulation of the virus to establish an endemic situation (Mariner et al., 2016).
Since the discovery of PPR, there have been many advances in the diagnosis of PPR in sheep and goats, and diagnosis protocols range from symptomatic diagnosis to virus isolation (Banyard et al., 2010). The current diagnosis is based on clinical symptoms, pathological lesions and precise identification of PPRV antigen or antibody in various molecular or serological tests in biological samples (Balamurugan, Hemadri, Gajendragad, Singh, & Rahman, 2014).
Virus isolation is the gold-standard for the diagnosis of PPR, but this is mostly impractical in the field (Banyard et al., 2010). As a morbillivirus, PPRV is antigenically similar to the viruses that cause rinderpest in cattle, measles in humans and distemper in dogs, but can be serologically distinguished by use of commercially available enzyme-linked immunoassay (ELISA) kits (Anderson & McKay, 1994).
In many countries, diagnosticians are moving towards the use of molecular techniques such as polymerase chain reaction (PCR) for early and specific diagnosis of PPR in sheep and goats (Kgotlele, Kasanga, Kusiluka, & Misinzo, 2014). Adaptation of a method of diagnosis often depends on local facilities and availability of resources (Banyard et al., 2010).
The first report of PPR was made in the Ivory Coast, West Africa, in 1942. Today, PPR is quite common in both Africa and parts of Asia, and is emerging as a threat to other continents such as Europe (Parida et al., 2016). In recent decades, several scholarly narrative reviews of the distribution of PPRV lineages based on nucleoprotein and fusion genes sequence analysis for particular geographical regions have been published (Banyard et al., 2010;Dilli, Geidam, & Egwu, 2011;Parida et al., 2016). A comprehensive pooled prevalence estimate of PPR has not been reported, but a wide range of prevalence estimates of PPR in sheep and goats have been reported in various regions Jaisree et al., 2017;Li et al., 2017). Reasons for the inconsistent prevalence estimates of PPR could include an endemic situation of PPRV in a particular geographic area, differences in methods used for identifying the disease, origin of samples, sampling strategy, year of study, study duration and species of animal studied. An overview of knowledge on the regional prevalence of PPR in sheep and goats will offer a better understanding of the distribution of the disease and its impacts on animal production, and will be useful in disease control. This study aims to use a systematic review and meta-analysis approach to estimate the regional prevalence of PPR in sheep and goats, and to evaluate the potential factors that contribute to the variability in the prevalence and distribution of disease.

| ME THODS
The study was conducted following PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines for systemic review and meta-analysis. The PRISMA 2009 checklist was used to ensure the inclusion of relevant information and maintain study standard (Appendix S1, Table S1).

| Search strategy
An optimized systemic search strategy was used to identify all published studies related to the prevalence of PPR in sheep and goats.
Published works were searched in four electronic web search en- Sheep OR Ram OR Ewe OR small ruminant). Search field option was selected as "All fields". A restriction was placed on the language of publication "English". Search terms and keywords were adjusted according to minor differences in syntax rules of four electronic databases. The reference management software EndNote X8 (Clarivate Analytics, Philadelphia, PA) was used to organize and remove duplicate articles between the search engines. The reference lists of extracted articles were also searched manually in triplicate for additional potential articles and to ensure that selected databases searches did not miss any reports.

| Selection of studies
Articles were selected for meta-analysis based on the following criteria: published between January 1969 and December 2018; fulltext article; published in English; any country of the world; studied population is sheep or goat or both; reported as animal level prevalence data; cross-sectional, case-control, longitudinal and cohort studies. Reasons for exclusion of articles were species other than sheep or goat, prevalence data not reported, case study or retroprospective study, flock with a history of vaccination, comparison of methods, experimental trial and articles in languages other than English.

| Quality of the studies
Studies selected for this meta-analysis were assessed for quality of reporting and selection for bias using a quality assessment checklist (Ahaduzzaman, 2019;Hoy et al., 2012). The checklist included nine parameters which have "yes" and "no" applicable option.

| Data extraction
The following data were extracted from each eligible study on a spreadsheet where possible: author, year, country, continent, study duration, species of animal, origin of sample, diagnostic technique, population size and number of positive samples. Overall, data from 243,864 animals (Sheep: 87,580 and goats: 156,284) from various geographical locations were analysed (Tables 1 and 2).

| Data analysis
All extracted data were transcribed and stored in a Microsoft Excel spreadsheet. Crude prevalence estimation was done by the number of positive animals divided by the total number of animals sampled TA B L E 1 Estimated pooled prevalence of PPR in sheep by world region and expressed as a percentage. Only the crude estimate of prevalence was used throughout, and the 95% confidence interval (CI).
The 95% CI of the mean was calculated using the standard formula for a proportion (p): where n is the studied population size. In the event that the higher limit of CI exceeded 100, then the value was settled to 100 to avoid >100% prevalence, or when the lower limit was lower than the positive value, then the value was settled to 0 to avoid negative prevalence. Data were analysed using the "metan" command of the STATA v.11.0 software (StataCorp LP, College Station, TX, USA). Heterogeneity across studies was assessed by chi-square test on Cochran"s Q statistic (represented as χ 2 and p-values) (Higgins & Thompson, 2002), which was interpreted by I 2 statistic value, considering the I 2 values of 25%, 50% and 75% as low, moderate and high heterogeneity respectively (Higgins & Thompson, 2002). Owing to the nature of the studies, there was substantial heterogeneity between studies; therefore, random effect meta-analysis was used for summary statistics.
Subgroup analysis was also conducted by world region, age, sex, origin of sample, methods of detection and study duration. Publication bias was assessed by Egger's test using two funnel plots, and the sources of funnel plots asymmetry were also tested to identify small study effects (Egger, Smith, Schneider, & Minder, 1997).  duplicates and screening of titles and abstracts, a total of 343 eligible articles were found, of which 148 articles were excluded due to following reasons: case report (n = 3); full-text not available (n = 9); individual prevalence data not available (n = 104); experimental trial (n = 12); article in a language other than English (n = 1); retro prospective study (n = 4); same dataset in different publication (n = 2) and others (n = 13). The list of excluded articles together with the causes for their exclusion is provided in Appendix S3, Texts S2. A total of 196 eligible articles were used for systemic review and meta-analysis (Abdalla, Majok, El Malik, & Ali, 2012;Temur, & Keleş, 2004;Yilmaz, 2016;Yousuf, Giasuddin, Islam, & Islam, 2015;Yousuf et al., 2017;Zahur et al., 2008Zahur et al., , 2009Zahur et al., , 2011Zahur et al., , 2014

| Prevalence estimates
The random effect meta-analysis showed that the pooled preva-  Table 3). The estimated pooled prevalence of PPR in sheep and goats by country is shown in Figure 2 and

| Source of heterogeneity
Heterogeneity in the prevalence of PPR in sheep and goats was due to six sources: world region (p < .0001), age (p < .0001), sex (p < .0001), origin of the sample (p < .0001), method of detection (p < .0001) and study duration (p < .0001). Overall, there was substantial heterogeneity (I 2 > 80%) in most pooled prevalence estimates ( Table 3).
The extent of publication bias in the included articles was mea-  Table S2).

| D ISCUSS I ON
This systemic review and meta-analysis summarises the prevalence of PPR in sheep and goats at regional level, based on a large population (n = 243,864). Articles included were from 34 countries from two continents, which allowed the analysis of reliable prevalence  (Parida et al., 2016). However, because of a lack of individual-level prevalence data or studies being in species other than sheep and goat, studies from Europe were not included in this meta-analysis.
Africa had a higher disease prevalence than Asia. Most of the individual studies showed a high prevalence; however prevalence estimates were low in only a few studies Cêtre-Sossah et al., 2016;. Reasons for the high prevalence of PPR in sheep and goats could include transboundary movement of infected animals with inadequate quarantine, the presence of hot and humid climatic conditions that favour disease epidemiology, lack of vaccination or vaccine administration monitoring, lack of awareness about PPR among backyard farmers, and limited funding for disease eradication in developing or underdeveloped countries.

TA B L E 4 Estimated pooled prevalence of PPR in sheep and goats in different countries
Moreover many studies included in this meta-analysis used serum sample or symptomatic diagnostic approaches to report PPR prevalence; such approaches can quickly reveal the status of a large population El-Rahim et al., 2010;Mahamat et al., 2018;Parvez et al., 2014).
The estimated pooled prevalence indicated that PPR is equally prevalent in both sheep and goats. Prevalence was higher in goats than sheep, but the difference was not statistically goats . Several individual studies reported that PPR is more prevalent in sheep than goats El-Yuguda et al., 2010;Enan et al., 2013) or more prevalent in goats than sheep Farougou et al., 2013;Fentie et al., 2018). Although there are biological differences between sheep and goats, higher prevalence in one F I G U R E 5 Forest plot of the prevalence estimates of Peste des petits ruminants (PPR) in goats amongst studies conducted in Africa | 823 AHADUZZAMAN species than another could be due to factors such as sampling process, richness or distribution of animal in a geographical area, management practices and strain of the virus. It is also possible that PPRV preferentially infects goats over sheep or vice-versa, depending on the endemic situation, and disease severity may also vary between species (Truong et al., 2014).
PPR was higher in adult animals than in young animals based on estimated pooled prevalence. These results agree with the F I G U R E 6 Forest plot of the prevalence estimates of Peste des petits ruminants (PPR) in sheep amongst studies conducted in Asia findings of many studies (Abubakar et al., , 2011(Abubakar et al., , 2017Acharya et al., 2018;Gari et al., 2017), but not with several others Bari et al., 2018;Bello et al., 2018). The higher prevalence in adults could be due to factors such as the higher likelihood of older animals being exposed to PPRV because of virus circulation, the foraging behaviour of adult animals and the decay of maternally derived F I G U R E 7 Forest plot of the prevalence estimates of Peste des petits ruminants (PPR) in goats amongst studies conducted in Asia antibody in older animals. It has been reported that PPRV is highly immunogenic, and animals remain seropositive for a long period, particularly in an endemic area (Acharya et al., 2018;. In contrast, higher prevalence in young animals could be due to malnutrition, less developed immune system and poor husbandry practices .
The estimated pooled prevalence of PPR was higher in female animals than in male animals. This could be due to breeding females being used for flock reproduction maintenance for a more extended period than males, while males are sold for meat at an early age (1-2 years) except those used for breeding purpose. Other factors could be a higher density of females than males in flocks, or physiological differences between females and males (e.g. females face some degree of stress as a result of production and reproduction).
The findings of this study are in agreement with previous findings Farougou et al., 2013;Mahamat et al., 2018).
Other studies found a higher prevalence in males, possibly due to a higher proportion of male animals in a flock particularly when the age of the studied animals was under two years .
The variation in the prevalence data in this report also depended on the source of the sample, the method of detection and the length of the study. Prevalence was higher in animals originating from abattoirs and farms, and in animals kept in mixed flocks. This may be due to there being a higher chance of having sick animals in abattoirs (because many farmers sell animals for slaughter during outbreaks) or to animals in abattoir coming from different regions and different farms; alternatively, it may be due to the method of detection in case of mixed flocks and farm-based studies, given that most of F I G U R E 8 Funnel plot for examination of publication bias of the prevalence estimates of Peste des petits ruminants (PPR) in sheep (a) and goat (b). Abbreviation: SE, standard error the studies were conducted using serology Rashid et al., 2008). Cross-transmission of PPRV between sheep and goats, or persistence of the virus in mixed flocks could be another contributing factor (Gari et al., 2017). Based on the method of detection, the prevalence was higher where the PCR-based method was used; this is due to PCR being more sensitive than other methods.
Studies conducted for less than six months showed a higher prevalence, and this could be due to a difference in study design.
Egger's test results using a linear regression approach and funnel plots asymmetry revealed strong evidence of publication bias.
However, the source of funnel plot asymmetry could additionally be due to true heterogeneity, location, data irregularity and artefacts or even to chance (Egger et al., 1997).

| Limitations
This study has several limitations. First, no reports on PPR prevalence in sheep and goats were found in continents other than Africa and Asia within the search range. Thus, it was not possible to obtain information on the prevalence from these regions. Second, age is an important factor to PPR, but there were not enough articles on age available for inclusion to carry out a multiple subgroup analysis; hence, the data are presented as young and adult. Third, the review excluded non-English articles, unpublished articles, retro perspective articles, method validation articles, results of experimental trials and case reports. Fourth, the genotype/sequencing data used to identify countries that are endemic for PPR or have PPR outbreaks (e.g. Bulgaria) have shown that PPRV is prevalent in many countries of the world, but it was not possible to include them in this study because of the lack of prevalence data,. Finally, heterogeneity in models was significant, suggesting that other factors that were not considered might have had substantial effects.

| CON CLUS IONS
The Africa-Asia pooled prevalence of PPR in sheep and goats estimated through this meta-analysis was high, and it varied between geographical regions and countries. The disease was found to be more prevalent in Africa than in Asia, and more prevalent in adult animals; thus, vaccination of young animals may prevent the disease. The findings suggest that screening tests for PPR, and effective preventive and eradication measures, should be routinely carried out in sheep and goat flocks in regions with a high disease prevalence, to control the outbreak and improve animal productivity. Further, there are regions where the virus is circulating, but no reports on prevalence estimates are available; therefore, epidemiological surveillance is needed for estimating disease status and eliminating the disease. Additionally, factors that contribute to the prevalence estimate heterogeneity should be handled appropriately in any survey to accurately estimate the true extent of PPR.

ACK N OWLED G EM ENT
The author would like to acknowledge the library facility of the University of New England, Australia which enabled access to literature databases and provided the technical supports. Thanks to Dr.
Hilary Cadman for language editing on the manuscript.

CO N FLI C T O F I NTE R E S T
The author declares that he has no competing interests.