Michael and Bundy (1997) used a newly assembled database on country-specific estimates of case prevalence (Michael et al., 1996), to construct the first maps of the spatial distribution of lymphatic filariasis case prevalences at both the global and regional levels (Figure 2.15A,B). A striking feature of the resulting maps was the high degree of geographical heterogeneity observed in the estimated country prevalences. In general, countries with bancroftian filariasis (the more important of the two disease forms) in Asia and South America appear to have lower prevalences compared to estimated country prevalences in the sub-Saharan African and Pacific Island regions (Figure 2.15A). The map for brugian filariasis (Figure 2.15B) appears to be relatively more homogeneous, although there is a slightly higher prevalence in the eastern regions of the distribution.

The authors investigated the apparent spatial heterogeneity for bancroftian filariasis distribution using simple statistical models for assessing the significance of area data (Cliff and Hagett, 1988). In particular, the approach of Poisson probability mapping was employed to construct maps of the statistical significance of the difference between disease risk in each study area and the overall risk averaged over the entire map. Such a mapping procedure not only stabilizes the individual prevalence rates for population size variations (which contributes to apparent heterogeneity), but may also provide a tool for highlighting truly anomalous areas (Bailey and Gatrell, 1995). The global probability map for bancroftian filariasis is displayed in Figure 2.16 and, although as expected the transformation of the country prevalences to a

Fig 2.15 Geographical distributions of bancroftian (A) and brugian (B) filariasis case prevalences based on the crude GBD estimates. Circles denote the corresponding prevalences (%) estimated for various Pacific islands and vary in size proportionately with the prevalence of each island. The figures in brackets indicate the number of countries

Fig 2.16 Global Poisson probability map for bancroftian filariasis case prevalences. The map shows ^¡>mean values, and may be interpreted by considering that there is a 'high probability' (p >0.90) that the prevalence estimated in each black area is higher than the mean global value (MGV); there is 'equivocal evidence' that the risk of each dark-shaded area is higher than the MGV (p=0.50-0.90) and that of each light shaded area is lower than the MGV (p = 0.10-0.50); and, finally, there is a 'high probability' that the risk of each medium shaded area is lower than the MGV (p<0.10). (Note: caste probabilities for all the other endemic Pacific Island countries lay between 0.50 and 0.90)

Fig 2.16 Global Poisson probability map for bancroftian filariasis case prevalences. The map shows ^¡>mean values, and may be interpreted by considering that there is a 'high probability' (p >0.90) that the prevalence estimated in each black area is higher than the mean global value (MGV); there is 'equivocal evidence' that the risk of each dark-shaded area is higher than the MGV (p=0.50-0.90) and that of each light shaded area is lower than the MGV (p = 0.10-0.50); and, finally, there is a 'high probability' that the risk of each medium shaded area is lower than the MGV (p<0.10). (Note: caste probabilities for all the other endemic Pacific Island countries lay between 0.50 and 0.90)

probability scale replaces the high spatial variation of the original map with a more homogeneous pattern in the 'between-country' distribution of cases, the results also confirm the impression from Figure 2.15A that the underlying case rate for the disease is not constant across the world. Instead, the case rates exhibit strong regional variations, with more countries in Africa and the Pacific Island region (not shown) with probabilities of infection and disease higher than the global mean rate, compared to countries in Asia or South America (Figure 2.16). This finding of a significant regional influence on spatial variation suggests that separate analytical maps based on regional mean rates will be required to identify anomalous or priority counties within each endemic region. They also argue for a geographically targeted strategy for filariasis control.

Was this article helpful?

## Post a comment