Epidemiological Study of Canine Leishmaniasis in Algeria and its Impact on the Human Visceral Leishmaniasis

Zoonotic visceral leishmaniasis (VL), caused by Leishmania infantum is a vector-borne disease, transmitted to humans by the bite of infected female phlebotomine sand flies. In absence of treatment the outcome is death. The disease is characterised by irregular fever bouts, loss of weight, hepatosplenomegaly and anaemia. It is endemic in the Indian sub-continent and in East Africa. In the Mediterranean region, it is present in rural areas, in villages, in mountain areas, but also in certain peri-urban zones, where the parasites survive in dogs and other mammals. In fact, domestic dogs are the principal reservoir of human VL and they also can develop a more generalised and fatal form of the disease, although more than half of the dogs remain asymptomatic, the individual outcome depending on both cellular and humoral immune responses.
Both symptomatic and asymptomatic dogs can infect the phlebotomine vectors, thereby enabling the transmission to other dogs or to people. This is the main reason why early detection and treatment of infected animals is the best way to reduce the risk of infection and, as such, forms an essential component in the prevention and control of the human disease.
Ideally, a so-called “gold standard” test is required to reliably and correctly determine the infectious status of an individual and/or the infection prevalence in a population in a certain region at a certain point in time. A gold-standard test provides an error-free classification of individuals as infected or infection-free. The discriminative power of a real-life test is quantified by measures of precision, more specifically the sensitivity and the specificity of the test. They vary in function of characteristics of the population studied.
In the majority of cases, a gold standard test does not exist. In order to overcome this lack, use is made of reference tests. The latter are often imperfect (sensitivity and specificity below one) and they allow estimation of a so-called apparent prevalence rather than the actual true prevalence.
A statistical solution to the problem of lack of gold standard test consists of a Bayesian approach, which allows us to combine data of a study with external (prior, independent) information, such as results from previous studies or expert opinion. This approach yields an estimate of the true prevalence, as well as estimates of the sensitivity and specificity of the tests used.
This thesis attempts to contribute towards a better understanding of the epidemiology of visceral leishmaniasis in Algeria. More specifically, our general objective (Chapter 3) was to estimate the true prevalence of canine leishmaniasis, in a situation where a diagnostic gold standard test was unavailable, and to describe the epidemiological curve and profile of visceral leishmaniasis in the country. With this in mind, the work is organised in nine chapters. The first chapter outlines the importance of leishmaniasis (respectively visceral and canine) in terms of public health and veterinary public health and provides an overview of the current knowledge base on canine leishmaniasis.
The second chapter is devoted to a description of the epidemiological methodology available for the estimation of the true prevalence and the test characteristics in the absence of gold standard test.
Chapter 4 describes an epidemiological study conducted in the capital city Algiers, comprising 462 dogs subdivided in four groups according to their function (stray dogs, guard dogs, farm dogs and pet dogs). A Bayesian approach was used to estimate the true prevalence of canine leishmaniasis in the four groups and to evaluate three serological tests: an indirect immunofluorescence test (IFAT), a direct agglutination test (DAT) and a commercial rapid test on gel, based on the protein rK39 (PaGIA). We were able to demonstrate variability in the test performance characteristics in the different groups, and this in terms of sensitivity and specificity. IFAT, although it is an imperfect test, is nevertheless the test of choice when the dog population targeted consists of stray dogs, but its specificity drops to a unsatisfactory 65.2% (IC: 60.2-73.5%) when applied to farm dogs: the fact of living together with other animals may induce cross-reactions. Stray dogs also yielded the highest estimate for the true prevalence (11.7%), probably a consequence of their constant living outside, which increases the risk of exposure to infected phlebotomine vectors.
Using the above results, a large-scale transversal survey was mounted in six towns of the littoral zone of Algeria (Chapter 5). Thus, a total of 2,184 farm dogs and guard dogs were sampled in two distinct periods using two serological tests: before and after the main vector season, using IFAT and DAT. Serial and parallel interpretation of the combined tests were compared to various other statistical methods in order to estimate the true prevalence and the sensitivity and specificity of the two tests. The latent class model of Hui and Walter based on conditional independence between the diagnostic tests was compared to a Bayesian model, which made it possible to include conditional dependency between tests, if required. The best-adjusted model estimated canine leishmaniasis prevalence between 11% and 38% with an increasing trend from west to east.
IFAT sensitivity varied between 86% and 88% in the different localities, whereas specificity ranged between 65% and 87%. DAT was invariably less sensitive than IFAT, but had a better specificity: between 80% and 95% in function of locality and season. The results highlighted once more the danger of considering IFAT to be a gold standard test for canine leishmaniasis.
Chapter 6 draws up an epidemiological profile of human zoonotic visceral leishmaniasis in Algeria, based on cases notified to the National Public Health Institute between 1998 and 2008. A total of 1,562 were found for this period. This translates into an incidence rate of 0.45 cases per 100,000 inhabitants, of which 81.4% were children between zero and four years of age. Splenomegaly, fever, pallor and pancytopenia were the principal clinical and biochemical signs. Antimonies are the first-line treatment drugs in paediatric wards and amphotericin B injections are the treatment of choice in adults. Severe under-declaration (less than one case in ten notified) was demonstrated thanks to a survey carried out in the paediatric, internal medicine and infectious diseases departments of the five university hospital centres in Algiers, which had an influx of patients from all four corners of the country. Results of both Chapter 5 and Chapter 6 revealed a concordance between human visceral leishmaniasis incidence rates and true prevalence of canine leishmaniasis, providing extra support for the hypothesis that dogs are the main reservoir of the disease.
Motivated by the results of the two surveys (Chapters 4 and 5), Chapter 7 describes a literature review of the performance of IFAT when used to estimate the canine leishmaniasis prevalence in the Mediterranean basin.
Chapters 8 and 9 present the general discussion and conclusions and recommendations of the overall contribution of this research effort. Even though this work only represent a small addition to knowledge body on visceral leishmaniasis, it allowed us to better describe and understand the epidemiology of visceral leishmaniasis in Algeria and to obtain true prevalence estimates of canine leishmaniasis in various towns in the littoral zone of the country. It showed us the importance of correctly identifying and defining the target population of surveys to obtain the best possible posterior estimates and to optimise control measures. This work also showed the efficacy of the application of appropriate statistical models in a Bayesian framework when confronted with the absence of a gold standard test.
Chapter 10 summarizes all references cited in the thesis.