Methods for estimation of associations between multiple species parasite infections

S.C.  HOWARD  a1 c1, C.A.  DONNELLY  a1 and M.-S.  CHAN  a1
a1 Wellcome Trust Centre for the Epidemiology of Infectious Disease, Department of Zoology, University of Oxford, South Parks Road, Oxford OX1 3FY


Human populations are often infected with more than one species of parasite, especially in developing countries where overall rates of parasitism are high. Infections with multiple parasite species may not necessarily be independent within an individual as physiological, immunological or ecological factors may result in positive or negative associations between infections with different parasite species. A general framework for estimation of these associations is presented. Data from over 215000 individuals are analysed and the associations between geohelminth (Ascaris lumbricoides, Trichuris trichiura and hookworm) and malaria species are investigated. A method is presented for analysing data from multiple communities and testing whether the associations in different communities are equal. Overall estimates of the associations between species are obtained for each country and continent where data were available. Associations between geohelminth species were, in general, found to be positive whilst both positive and negative associations were found between the different Plasmodium species. There was evidence for significant geographical heterogeneity between the associations. A method for using these parameter estimates to predict the distribution of multiple infections when only marginal prevalence data are available is described and demonstrated.

(Received May 27 2000)
(Revised July 20 2000)
(Accepted July 20 2000)

Key Words: multiple species; log-linear; geohelminths; Plasmodium spp.

c1 Corresponding author: Stroke Prevention Unit, Department of Clinical Neurology, Radcliffe Infirmary, Woodstock Road, Oxford OX2 6HE. Tel: +01865 224152. E-mail: