Peste des Petits Ruminants (PPR) is a severe and highly infectious viral disease of small ruminants. The PPR virus (PPRV) belongs to the genus Morbillivirus in the family Paramyxoviridae. It is closely related to the rinderpest virus of bovines and buffaloes, distemper virus of dogs and other wild carnivores, human measles virus and Morbilliviruses of marine mammals [1–4]. In small ruminants, infection by PPRV is characterized by sudden depression, fever, nasal and ocular discharge, diarrhoea and occasionally death. Morbidity in the range of 10 to 80% and mortality proportions from 0 to 90% have been reported. The wide range of reported values is likely to be influenced by differences between species (sheep or goats), production systems and levels of natural or acquired immunity [5–10].

PPR was first described in West Africa in 1942 [11]. Nowadays the disease is recognized as responsible for mortality and morbidity across most of the sub-Saharan African countries situated north of the equator, in the Arabian Peninsula, in India and in numerous other countries in Asia [6, 12–14]. Although nationwide serosurveys have been conducted in countries such as the sultanate of Oman, Turkey, Jordan and India, information on the frequency and distribution of PPR is often lacking when control or eradication campaigns are initiated [15–18]. Control of PPR in endemic areas relies mainly in vaccination [19, 20]. In 1989 a homologous vaccine that induces lifelong immunity in both sheep and goats was developed [6, 21–23]. The vaccine is innocuous on pregnant sheep and goats at any stage of gestation and induces the production of colostral anti-PPR antibodies that have been found in kids up to 3 months old [6, 23].

Ethiopia has the most important livestock population in Africa and is ranked 9^th in the world [24]. The livestock sub sector accounts for 40% of the agricultural gross domestic product (GDP) and 20% of the total GDP (Aklilu Y. An audit of livestock marketing status in Kenya, Ethiopia and Sudan. PACE/OUA/IBAR, 2002) without considering the livestock contribution in terms of traction power, fertilizing and mean of transport. Sheep and goat populations are estimated to be 20.7 million and 16.4 million respectively [25]. Sheep and goats contribute 25% of the meat domestically consumed with a production surplus mainly being exported as live animals [26, 27]. Both species also contribute 50% of the domestic needs in wool, about 40% of skins and 92% of the value of hides and skin exported [28]. The annual production of sheep and goat meat is estimated as 56 560 and 28 650 tonnes respectively [24]. PPR was clinically suspected for the first time in Ethiopia in 1977 in a goat herd in the Afar region, East of the country [9, 29]. Clinical and serological evidence of its presence has been reported by Taylor (1984) and later confirmed in 1991 with cDNA probe in lymph nodes and spleen specimens collected from an outbreak in a holding near Addis Ababa [29]. During the nineties, several small serological surveys were conducted, mainly east of an imaginary line that would run parallel to the Rift valley and pass through Addis Ababa. In 1994 Roger and Bereket (CIRAD-EMVT report n°96006, Montpellier, 1996) found seroprevalences of up to 33% in sheep and 67% in goats near selected urban areas. In 1996 Gelagay found that 14.6% of sheep sampled along 4 roads from Debre Berhan to Addis Ababa were seropositive [30]. In 1997 Yayerade found up to 100% of seropositive individuals in groups of adult male sheep and animals that survived suspected outbreaks. Although these studies provide very limited and potentially biased information about the frequency and distribution of PPR in Ethiopia, they clearly suggest that the virus has been circulating extensively among the small ruminant population of Ethiopia during the nineties. Based on the reported morbidity and mortality of the infection and the size and structure of the small ruminant sector it is likely that PPR became one of the most economically important livestock diseases in the country [12, 31].

In 1999, a serological survey on PPR was conducted in Ethiopia with the aim of informing a subsequent vaccination campaign which would be the first large scale vaccination campaign against PPR in the country. As part of the survey, a total of 13 753 sheep and goats were sampled. To our knowledge, this is the largest serological survey on PPR ever conducted in Africa. The objective of this paper is to describe the results of this survey and discuss its practical implications.

Although estimates have to be interpreted with caution because it had not been possible to ensure that random selection was used at all sampling stages, the results indicate that PPR has been circulating in most of the country before large vaccination campaigns were implemented. Lack of large scale vaccination campaigns before the survey was conducted suggests that our seroprevalence estimates are likely to reflect infection [31]. Given the sensitivity and specificity of the test our results are likely to overestimate, slightly, the true proportion of seropositive animals [34, 35]. On the other hand, by using the results of an imperfect test as indicators of true infection we are probably underestimating the true values of rho [36]. Despite an overall frequency of 6.4%, the seroprevalence of PPR was above 50% in some weredas. PPRV circulation before 2000 has been heterogeneous: areas of low altitudes appear to have suffered more from infection than areas of highlands. Reasons for this may be related to different production systems with exchanges and movements in areas of lowlands being more frequent and involving larger numbers of animals. In Ethiopia small ruminants mainly thrive on free range pasture lands, shrubs and forest grounds. Agro climatic conditions influence the availability of these resources and the movement of animals becomes necessary in order to ensure the provision of fodder and water. This is particularly important during the dry season and in low altitude areas where resources are scarce. In addition, animals are exchanged between households and flocks as a result of social practices and changes in economic conditions that exhibit seasonal patterns. The seasonality of animal movements could partly explain the occurrence of the disease in Ethiopia mainly between the months of March and June [7, 17, 31].

Although the overall seroprevalence of PPR in Ethiopia appears to be low compared, for example, to the 22.4% reported in Turkey and 33% in India, it is difficult to draw any conclusions because of the differences in sampling procedures in the different studies that affect their representativeness [16, 17]. However, a common feature described by the respective authors are heterogeneities, possibly related to agro-climatic and socioeconomic factors.

Age appears to be a risk factor for seropositive status, and its linear effect suggests that PPRV is highly immunogenic, naturally infected animals remaining positive for a long time.

The intra-kebelle correlation coefficient was found to be very low in most wereda with a small number of them showing high values. Differences in biological factors probably explain this variability. One hypothesis could be the past or recent circulation of PPRV reflected by a low or a high value of ρ, respectively, along with a low or high seroprevalence. Assuming that a high inhibition percentage could reflect recent infection, the strong correlation between ρ and inhibition percentage would be consistent with the interpretation of high ρ as being indicative of current or epidemic presence of the virus in a few kebelles within the wereda. This correlation then diminishes with time, diluting itself in a wereda as a consequence of a relatively rapid turnover of small ruminants (3 years), PPRV being highly immunogenic and that these are serological results. Considering that PPR is as a highly contagious disease and that both the within-and between-kebelle spread of infection determine ρ, the low value in certain weredas could also be attributed to weredas where animals of different kebelles mix a lot at market points or at water sources. The absence of an obvious spatial pattern in the distribution of ρ may reflect that spread of the disease has mainly occurred within individual wereda as opposed to large scale outbreaks involving several contiguous wereda.

Probabilistic sampling is a challenging task in a country with an infrastructure such as Ethiopia, since large areas have to be covered which are not easily accessible. Moreover, sampling frames of lower administrative units are often not available at central level. Under these circumstances, multi-stage sampling strategies such as the one used in the current study are the best option. Despite the random selection of weredas within regions, kebeles within weredas and villages within kebeles there is still potential for bias influencing our results due to non-random selection of regions and animals within villages. Although the large variation of values of ρ highlights the limitation of using a summary measure of ρ for a whole country as a basis for a sampling design, our estimates can inform the design effects needed to adjust for cluster sampling in future surveys on PPR in regions with similar production systems. As an example, if we consider the median ρ = 0,029, the standard sample size calculations using simple random sampling with 95% confidence interval for an estimated prevalence of 5% and an accepted error of 1% needs to be inflated by a factor of: D = 1 + 0,029(n-1), n being the average cluster size and D the design effect. That accommodates for the lack of independence between small ruminants belonging to a given kebelle [37].

Thus, to design a seroprevalence study at wereda level in Ethiopia, the clustering effect of the kebelles implies the sample size has to be increased by a factor of 1.26 if 10 small ruminants are to be sampled per kebelle, 1.55 if 20 small ruminants are to be sampled, 2.42 if 50 are sampled and 3.87 if a hundred small ruminants are to be sampled in each selected kebelle. Our findings are in agreement with other published values for ρ and D related to viral diseases or vector-borne infections. Thus the majority of ρ reported lay below 0.20 with widely varying estimates for highly contagious viral infections as Infectious bovine rhinotracheitis (IBR) [38]. To our knowledge no specific reference was available until now for peste des petits ruminants.

Our study shows that PPR has extensively circulated across Ethiopia, but that there is large variation between regions and weredas. Although in most weredas there is a small variation between kebelles in some of them there are large differences that indicate the virus has only been introduced recently among some kebelles of the wereda, if our interpretation of high intracluster correlation coefficients as indicative of more recent introduction is valid, PPRV has been more recently circulating in the North of the country. The results of our study indicate that further research is needed to investigate the association of the presence of disease with the management practices in place. These findings are also important to direct future studies in other countries where PPR is of importance.