J. C. Mariner and A. Catley
African Union/Interafrican Bureau for Animal Resources/Community Animal Health and Participatory Epidemiology Project, the PACE Epidemiology Unit and RDP Livestock Services
Introduction
Now that rinderpest is largely controlled in Africa, the attention of animal health authorities has again turned to contagious bovine pleuropneumonia (CBPP). The disease was controlled by vaccination, movement control and slaughter programmes up until the mid-1980s. However, general economic decline led to a reduction in animal health budgets and a resurgence of CBPP in the late 1980s and 1990s. CBPP is now endemic in most pastoral areas of East, Central and West Africa, and is spreading in southern Africa. Many transhumant communities identify CBPP as a major priority for attention. There has been little change in terms of the tools available to control CBPP over the last 30 years, but social, economic and political realities have changed dramatically. Control actions that worked in the past may no longer be feasible.
This report describes the results of a study to model the transmission of endemic contagious bovine pleuropneumonia (CBPP) in transhumant production systems. The objectives of the study were to construct a model of CBPP transmission and then to evaluate the impact of alternative control strategies in an effort to help guide decision-makers in the formulation of realistic recommendations to national governments for the control of CBPP in light of current socio-economic conditions.
The modelling study built upon a previous participatory data collection study (Mariner et al., 2003). Field data and the expert opinion of livestock owners on the behaviour of CBPP in their herds were utilized as a foundation for the modelling. The literature on prevalence, vaccine efficacy, pathology, diagnostic tests and epidemiology has been investigated in detail. The parameter estimates and model structure were developed using data and insights derived participatory epidemiology, serologic studies and an evidence-based literature review. In this way, the analysis was solidly grounded in the reality of the field yet incorporated all available data.
The model has a stochastic, state-transition design and was constructed using @Risk Software. At present, the model includes three inter-linked sub-populations and disease transmission was modelled in both homogeneous and structured populations. Using serological data, the basic reproduction numbers for CBPP in southern Sudan was estimated as 3.2 to 4.6 with a most likely value of 4.1.
Participatory Epidemiology
Participatory epidemiology (PE) is the use of participatory methods to collect epidemiologic data (Mariner, 2000). The participatory approach utilizes and compares all available information through a process called triangulation. Triangulation refers to the confirmation of information using multiple methods and multiple sources. Participatory epidemiology makes full use of sampling, laboratory testing and analytical techniques. Participatory studies provide the proper background for the design of statistical or mathematical studies and the contextual information that is essential to the correct interpretation of laboratory results.
Pastoralists have a very well developed knowledge of the clinical presentation, epidemiologic patterns and principal pathology of CBPP. In general, livestock owners are aware of the range of clinical presentations of CBPP from mild to severe or acute to chronic. They report that individual cases can last from a few days to up to 12 months. Prolonged clinical cases are often perceived as intermittently episodic. Fully recovered animals are reported to be immune for life. In transhumant communities, a pronounced seasonality in clinical incidence was reported that could be related to grazing movements leading to changes in mixing patterns and contact rates.
Pastoralists clearly recognize that the disease is contagious and that cattle movement, livestock exchanges and contact are major risk factors for the disease. However, the respondents viewed the relative importance of movement in the production systems and the positive socio-economic contributions of cattle exchange to be relatively more important to their pastoral livelihoods than the impact of CBPP. They were risks worth taking relative to the essential benefits entailed.
Livestock owners were also aware of presence of inapparent infection and reported that the source of infection in clinical cases that developed after a period of absence of clinical cases was from within their own herds.
In regard to control interventions, livestock owners and field veterinarians reported treatment with antibiotics had a clear positive effect. For the most part, treatment regimes described could be described as sub-optimal. On the other hand, livestock owners reported that the mass vaccination programs, as currently practiced with available resources, had limited or no impact on the incidence of CBPP.
Summary of the Spatially Heterogeneous Model for CBPP Transmission
The model utilized a stochastic state-transition approach with an open population structure. The final model was spatially heterogeneous in that it incorporated three inter-linked sub-populations. One population was treated as a reference population and the results reported relate to the status of this reference population at the termination of the model. The model incorporated seasonal forcing of contact rates. Vaccination was modelled as pulses delivered in one single time step that could be repeated at the user’s discretion. All simulations were allowed to run for six-years.
The model incorporated six principal states, susceptible (S), exposed (E), infectious (I), recovered (R), vaccinated (V) and chronically infected (Q). Vaccination and naturally recovered immunity were modelled as separate states due to the difference in the quality and duration of immunity resulting from vaccination and natural infection. There were two recovery routes for infected animals. The first was full recovery directly to the recovered state (R). The second pathway was recovery through the chronically infected state (Q). This was a process involving sequestration where animals developed encapsulated, infected but non-infectious lesions. The majority of these animals progressed to the recovered state (R), but in a small percentage of cases relapsed, infection was re-activated and the animals re-entered the infectious state (I).
The transmission rate parameters used in the model were derived from serologic studies conducted as part of the research and information available in the literature. The basic reproductive number for CBPP transmission in pastoral communities of southern Sudan was determined from the average age of infection. Model validation consisted of comparison of model predictions of prevalence with prevalence measurements derived from serologic studies.
The effective within population contact rate was based on estimates of the basic reproductive rate. The between population effective contact rate was set as a percentage of the within population contact rate. Unless otherwise specified, between herd contact was set to 10 per cent of the within herd contact.
Principal Results
The model was used to derive estimates of the critical community size for the maintenance of CBPP infection in both heterogeneous populations and single isolated herds. It was found that a heterogeneous population of three inter-liked herds with as few as 50 head per herd were capable of supporting persistent infection over a period six years. On the other hand, in a single isolated herd, more than 300 head was required to support indefinite transmission at equivalent levels of herd prevalence. This points to the efficacy of quarantine and movement control in regard to CBPP control.
A sensitivity analysis was conducted using various levels of inter-herd contact. It was found that when the between herd contact rate was set to 1 per cent of the within herd contact rate, the disease remained endemic in the reference in herds in more than 46 per cent of the iterations of the model. This indicated that only strictly enforced quarantines could be effective. However, the results of the participatory data collection indicated that this level of movement control was not a realistic option under pastoral conditions.
Mass vaccination was modelled for a range of vaccination efficacy (50 to 80 per cent), an 80 per cent level of population coverage and 80 per cent efficiency of vaccination. The average duration of vaccinal immunity was set to 3 years. It was shown that even in 5-year annual or biannual campaigns, vaccination alone was unable to eliminate infection from an important percentage of herds. For example, after 5 years of biannual mass vaccination with a vaccine of 70-80 per cent efficacy the disease persisted in the population in more than 20 per cent of the model iterations.
Elective control methods based on vaccination were modelled by applying vaccination to only one of three sub-populations in the heterogeneous model. The other two sub-populations were left without intervention and served as a potential source for the reintroduction of infection to the vaccinated population. In these simulations, the herd level prevalence of infection among vaccinated herds remained essentially unchanged but the individual animal mortality within the vaccinated herd was substantially reduced. Where 178 deaths could be expected over a six-year period in an unvaccinated herd of 500 head, expected losses were reduced to 124 deaths in a vaccinated herd if a 5-year program of annual vaccination was applied electively. Thus, the livestock owner was able to capture most of the potential benefit in terms of the reduction of losses that could result from mass vaccination in an elective vaccination program. This suggests that elective control is a viable option that has the advantage of being driven by private investment. Such an approach would require the liberalization of CBPP vaccine to empower individual livestock owners and private service providers to embark on elective programs.
The potential impact of treatment was modelled by reducing the average infection period by 25 and 50 per cent and observing the impact of the change on persistence on infection in the reference herd. It was found that a halving of the average infectious period could reduce herd level prevalence at the end of the modelling period from 75.4 to 33.2 per cent. Mortality in a 500 head herd was reduced from 178 to 64 head over a 6-year period. Thus, a reliable treatment regime could have impact as great as or even superior to currently available vaccines.
A combined program of treatment and vaccination was evaluated. Such a program would consist of treatment of all diseased animals and vaccination of the remainder of the herd. A five year program of biannual vaccination combined with treatment of cases was found to be the only option modelled that approached eradication. At the termination of the model, the reference herds remained infected in only 0.4 per cent of the iterations.
The relation between mortality and prevalence of infection was investigated in populations of 10,000 head. When a case fatality rate of 33.2 per cent was incorporated in the model, the predicted prevalence of infection was 6.3 per cent and the total mortality was 1,338 head over the life of the model. On the other hand when a case fatality rate of 4.7 per cent was used, the predicted prevalence of infection was 9.0 per cent and the total mortality loss was 235 head. This simulation illustrates the point that prevalence and impact are not necessarily positively correlated as high mortality rates can decrease the duration of infection leading to a lower prevalence of infection. Prominent epidemiological texts describe this phenomena as a key drawback to the use of prevalence surveys (Martin et al., 1987; Rothman and Greenland, 1998).
Conclusions and Recommendations:
Based upon the results of the field investigations, modelling work and literature review, the following recommendations can be made relative to CBPP control:
mass vaccination alone with the currently available T1 44 vaccine is unlikely to eradicate CBPP even when applied biannually over a 5-year period;
a regular programme of mass vaccination can have a major impact on morbidity and mortality losses;
a proven treatment regime would have considerable impact on both the prevalence of CBPP and the morbidity, mortality and production losses associated with CBPP infection;
a combined vaccination and treatment programme offers potentially greater impact (up to 44 per cent over annual vaccination campaigns alone) than either approach in isolation;
individual livestock owners who adopted a regular programme of elective control can capture most of the benefit in terms of reduced disease morbidity, mortality and production losses that would result from a mass vaccination programme even if in-contact herds do not practice vaccination;
levels of movement control consistent with sustainable pastoral livelihoods are unlikely to have a major impact on CBPP prevalence, and in the current socio-economic climate movement control is unlikely to contribute to CBPP eradication in endemic areas.
The following recommendations can be made regarding CBPP surveillance and economic analysis:
abattoir surveillance for CBPP lesions is a useful technique for the detection of CBPP in an area but should not be used to estimate prevalence due to problems of sensitivity and bias;
prevalence surveys are an insufficient approach to the assessment of CBPP impact as prevalence and mortality impact are not necessarily positively correlated;
broad-based impact assessments conducted at the community level that incorporate participatory approaches and purposive use of laboratory testing are more likely to correctly characterize CBPP morbidity, mortality and production losses than random sero-survey in isolation. Impact assessments can also serve as a reliable basis for the identification of realistic control options and economic analysis that includes livelihood effects.
Given the paucity of socio-economically acceptable tools for CBPP control, the near universal use of antibiotics for treatment of CBPP and the potential impact of well designed treatment regimes, research to validate treatment regimes should be conducted without delay. This neglected area warrants at least as much investment as vaccine development has enjoyed in the past. The objective of the research should be the identification of proven, affordable treatment schedules that can be incorporated into training messages for both veterinary staff and community animal health workers. Provided an effective regime is documented, an approach that combines treatment of clinical cases with vaccination of animals at risk should be developed.
Many national governments lack the means to carry out mass vaccination and international donors are highly unlikely to support an expensive and open-ended control programme. Due to funding constraints, the present policy of government-sponsored mass vaccination actually restrict livestock owner access to vaccine. Strategy guidelines need to be developed focussing on elective vaccination delivered by private and community-based service providers on a strictly commercial basis. Such an approach will require policy reforms that create adequate private sector access to regulated supplies of CBPP vaccine. This will enable livestock owners to develop reliable CBPP herd health programmes in consultation with their veterinarians and community-based service providers.
References
Mariner, J. C. (2000). Manual on Participatory Epidemiology. Food and Agriculture Organization, Rome, pp. 81.
Mariner, J. C., Aluma Araba, and Makungu S. (2003). Consultancy on the Dynamics of CBPP Endemism and the Development of Effective Control/Eradication Strategies for Pastoral Communities: Final Data Collection Report. Nairobi, The Community Animal Health and Participatory Epidemiology Unit of the African Union Interafrican Bureau for Animal Resources.
Martin, S.W., Meek, A.H. and Willeberg, P. (1987). Veterinary Epidemiology: Principles and Methods. Iowa State University Press, Ames, pp. 343.
Rothman, K.J. and Greenland, S. (1998). Modern Epidemiology. Lippincott - Raven, Philadelphia, pp. 738.
