Gram02c903

Planning Disease Control in Cameroon, West Africa Simon Brooker, Simon I. Hay, Louis-Albert Tchuem Tchuente 1995a). Praziquantel is used to treat the schistosomes (Schisto- soma haematobium and S. mansoni), and intestinal nema- Human helminth infections (intestinal nematode infections todes (Ascaris lumbricoides, Trichuris trichiura, hookworm) such as Ascaris lumbricoides, Trichuris trichiura, and hook- are treated by the benzimidazole drugs, albendazole and meb- worm, and schistosome infections such as Schistosoma endazole (WHO, 1995b). Studies have shown that these treat- haematobium and S. mansoni) affect more than a quarter of ments can be safely and effectively combined (Savioli et al., the world’s population, with potential consequences for the 1997; Olds et al., 1999), and the WHO recommends joint deliv- health and nutritional and educational development of infected ery in areas where both groups of parasites occur (WHO, individuals. The advent of broad-spectrum anthelminthic 1995a). It has been suggested that the overlap in the geographi- drugs that are cheap, safe, and simple to deliver has meant cal distribution of each group of species is sufficiently large to that control has become a viable option for many countries. justify combined treatment (Bundy et al., 1991), but more de- Because helminth infections patterns are highly heterogeneous, tailed analysis suggests uneven and often non-overlapping dis- methods to identify priority areas for intervention against tributions within countries (Brooker et al., 1999). This indi- intestinal nematode and schistosome will enhance the efficacy cates a more refined approach to combined control is required, of control. This paper describes the use of NOAA-AVHRR data whereby target communities are identified separately for inter- to develop logistic regression models that predict the prob- vention against schistosomes and intestinal nematodes, and ability of infection prevalence greater than 50 percent, and drugs are distributed according to local needs, thus reducing thus warrant mass treatment for intestinal nematodes and delivery costs and the prospect of drug resistance.
schistosomes, according to WHO’s criteria. Moreover, by In an effort to better understand the distribution of species, overlaying the resulting risk maps on population surfaces, it geographic information systems (GIS) are increasingly being is possible to estimate the school-aged population size used to collate and map available helminth survey data avail- requiring mass treatment and also provide an estimate of able from the formal and “grey” literature (Brooker et al., 2000a). Such information on the distribution of infection willbe central to successfully addressing the key operational ques- tions of reliably estimating the target population numbers at More than fifty years ago, a seminal paper entitled “This risk (Brooker et al., 2000b), stratifying areas by prevalence to Wormy World” (Stoll, 1947) showed that helminth infections, prioritize areas for control, and estimating overall drug needs including schistosome and intestinal nematode species, were and costs. Although there is comprehensive information on among the most common of human infections. Today, these helminth distributions in some African countries, empirical infections still affect more than a quarter of the world’s popula- survey data are available for only a third of administrative dis- tion (Chan et al., 1994; Bundy, 1997), with potential conse- tricts across the continent (Brooker et al., 2000a).
quences for children’s physical and intellectual development To help fill the gap in empirical data, remotely sensed (RS) (Stephenson, 1987; Watkins and Pollitt, 1997).
satellite sensor data and interpolated meteorological surfaces The advent of broad-spectrum anthelminthic drugs that are being used to predict the distributions of a variety of infec- are cheap, safe, and simple to deliver has meant helminth con- tious diseases (Malone et al., 1997; Hay et al., 2000; Rogers, trol has become a viable option for many countries. The World 2000; Lindsay and Thomas, 2000; Malone et al., 2001). For hel- Health Organization (WHO) presently recommends mass an- minth species, years of field studies have documented the thelminthic treatment in areas where infection prevalence (pro- influence of climate and environmental variables on the distri- portion of community infected) is 50 percent or greater (WHO, bution of helminth infections (Appleton, 1978; Brown, 1994;Crompton, 1994), and RS-derived environmental variables andmeteorological variables are of potential use in predicting the S. Brooker is with the Department of Infectious Disease Epide- occurrence of significant transmission (Brooker and Michael, miology, Imperial College School of Medicine, Norfolk Place, 2000). The present study uses environmental data derived London WG 1PG, United Kingdom ([email protected]).
from meteorological satellite sensors and interpolated meteoro- S. Hay is with the Trypanosomiasis and Land Use in Africa(TALA) Research Group, Department of Zoology, Universityof Oxford, South Parks Road, Oxford OX1 3PS, United Kingdom.
Photogrammetric Engineering & Remote Sensing L.-A. Tchuem Tchuente´ is with the Centre for Schistosomiasis & Vol. 68, No. 2, February 2002, pp. 175–179.
Parasitology, P.O. Box 7244, Yaounde´, Cameroon.
R. Ratard is with the Louisiana Department of Health and Hos-pitals, Office of Public Health, 1201 Capitol Access Road, P.O.
᭧ 2002 American Society for Photogrammetry Box 629, Baton Rouge, LA 70821-0629.
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING logical data to model the distribution of schistosome and intes- Logistic regression models were developed to identify signifi- tinal nematode infections in Cameroon. These distributions are cant environmental variables affecting the transmission of then used to quantify the population requiring treatment and infection. A potential problem in developing regression mod- to estimate the financial costs of school-based programs using els using environmental vari-ables is that many are highly inter- single species or combined control options.
correlated so that it is difficult to separate the effects of theindependent variables statistically (Morgenstern, 1998). Toreduce the dimensionality of these collinear variables, we first selected those variables likely to have greater biological signifi- cance on infection transmission (Brooker and Michael, 2000).
Prevalence data were collected during a nation-wide survey of Second, the remaining variables were added to the models in a helminth infection in 1985–1987 originally aggregated at the stepwise fashion, and the statistical fits of alternative models district level (Ratard et al., 1990; Ratard et al., 1991). These data were compared using the residual deviance of models includ- are stratified at the school level here because schoolchildren ing and excluding correlated variables using a ␹2 distribution are the primary targets for treatment and the educational infra- (Venables and Ripley, 1999). Analysis was done using S-Plus structure is usually used to deliver treatment. The original 4.5 Professional Release 2 (Math Soft, Seattle, Washington).
study was designed to provide data on the nationwide distribu- The best-fit logistic regression models were then used to tion of helminth infections using a stratified, random-cluster map the probability of infection prevalence being greater than sampling procedure with the primary school as the basic sam- 50 percent using Idrisi Version 2 (The Idrisi Project, Worcester, pling unit. In the north of the country where school enrollment Massachusetts). To define whether a district would be a prior- was low, all children from the appropriate age group living in ity area for control, we have used an arbitrary criteria based on the community were invited to participate (Ratard et al., 1990).
whether the average logistic regression probability is greater Urine and stool samples were examined by sedimentation and than 0.5 within a district. On this basis, the number of school- the Kato-Katz thick smear technique, respectively. The loca- aged children who would receive mass treatment for intestinal tion of schools was obtained by transcribing coordinates from nematodes was estimated. The population size for praziquantel 1:25,000-scale maps used in the original survey. Data on intes- and albendazole treatment was calculated by overlaying the tinal nematode species (Ascaris lumbricoides and Trichuris predictive maps of infection prevalence on a population map.
trichiura) for 18,260 school children aged 10 to 19 years in 402 For albendazole we used a combined estimate of infection pre- schools and data on schistosome species (Schistosoma haema- diction for either A. lumbricoides or T. trichiura.
tobium) for 19,524 children in 303 schools were collected.
Detailed prospective cost analyses have been conducted for Land Surface Temperature (LST) and the normalized difference
school-based anthelmintic programs in Ghana and Tanzania vegetation index (NDVI) information were derived from the
(Partnership for Child Development, 1999a). The cost of mass Advanced Very High Resolution Radiometer (AVHRR) on board distribution of a single dose of albendazole for intestinal nema- the National Oceanic and Atmospheric Administration’s todes to schoolchildren by their teachers was US$0.03 in both (NOAA) polar-orbiting meteorological satellites (Cracknell, countries. The cost of delivering praziquantel for schisto- 1997) using Price (1984) and Tucker (1979) procedures, respec- somes—which required targeting schools by a questionnaire, tively. Daily data at 8- by 8-km spatial resolution were first pro- and required a calculation to determine the dose based on the cessed for the period 1985 through 1998 to exclude unreliable height of the child—was US$0.67 in Ghana and US$0.21 in pixels due to extreme sun and sensor viewing angles and cloud Tanzania. The figures for Tanzania and Ghana were used to contamination (see Hay and Lennon, 1999). Single monthly estimate the lower and upper costs of implementing a school- images were then maximum-value composited (Holben, 1986).
based helminth control program in Cameroon.
Minimum, mean, and maximum values of these data wereextracted for each pixel that corresponded to the location of the parasitological surveys. Image processing was performed using the Earth Resources Data Analysis System (ERDAS) Imag- A number of different logistic regression models were fit to the ine 8.4௣ (ERDAS, Inc., Atlanta, Georgia).
data, and residual deviances were compared to identify thebest-fit models. The variables available to the regression analy- sis were mean, minimum, and maximum LST and NDVI; total Interpolated rainfall surfaces were taken from the Spatial Char- annual rainfall; and altitude. Studies show that maximum tem- acterization Tool (Corbett and O’Brien, 1997), and an interpo- perature is an important variable in determining helminth dis- lated digital elevation model (DEM) of Africa was obtained from tribution because of the effect of heat and low humidity on the the Global Land Information System (GLIS) of the United States embryonation, development, and survival of free-living infec- Geological Survey (EROS Data Center, 1996).
tive stages and intermediate hosts (Brooker and Michael, 2000).
Consequently, this variable was entered into the regression model first; next, minimum and mean LST were included and District population data were derived from a 1990 national the additional model improvement was assessed. Added next population forecast based on the 1987 national census (Deich- to the model analysis was NDVI (minimum, maximum, and mann, 1996), and were projected to 2001 using annual specific mean), rainfall, and altitude. The derived species-specific growth rates obtained from the United States Census Bureau models (Table 1) indicate (1) the importance of maximum LST, (2) the influence of rainfall, and (3) the influence of NDVI.
The results for A. lumbricoides and T. trichiura indicate a negative effect of maximum LST. For helminth species, temper- To examine the relationship between environmental variables ature is a density-independent factor effecting parasite trans- and the need for mass treatment, schools were classified as mission, as measured by the basic reproductive number (R0) having prevalence above or below 50 percent, WHO’s treatment (Anderson and May, 1991; Brooker and Michael, 2000), and threshold (WHO recommends that mass treatment is warranted thus observed patterns of infection prevalence. As temperature if the prevalence in a school exceeds 50 percent infection).
increases, transmission and infection prevalence decrease.
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING REGRESSION COEFFICIENTS DESCRIBING THE LOGISTIC REGRESSION no districts would warrant mass treatment at the 50 percent MODELS. LST ϭ LAND SURFACE TEMPERATURE; NDVI ϭ NORMALIZED threshold. However, because there will be heterogeneity of prevalence within districts and there should be flexibility inthe treatment thresholds to suit local needs, the analysis was re- run using a 20 percent prevalence threshold. On this basis, we estimate that 1.8 million school-aged children in nine districts would receive mass treatment with praziquantel. For S. haematobium, an effective approach to help locate high-risk communities/schools requiring mass treatment within at-risk areas has been the use of blood in urine questionnaires (Red Urine Group, 1995; Partnership for Child Development, 1999b). To identify priority areas for questionnaire surveys, the present broad-scale ecological predictions can provide rele- Estimates of treatment costs were developed based on the predicted target population size. Using the Tanzania cost data as a lower estimate and the Ghana cost data as an upper esti-mate, we suggest that the cost of control for A. lumbricoidesand T. trichiura (using albendazole) would be US$ 0.18 millionand for S. haematobium (using praziquantel) would be In contrast to the models for A. lumbricoides and T. trichi- ura, maximum LST has a positive effect on S. haematobium. InCameroon, the predominant snail species are Bulinus senega- lensis in the north of the country and Bu. globosus and Bu. These results indicate that RS and meteorological data offer the truncatus in the south (Wright, 1959; Greer et al., 1990). The opportunity to investigate the distribution of intestinal nema- former is found principally in semi-permanent water bodies tode and schistosome infection and some of the ecological fac- and can survive the dry season by aestivation. By contrast, the tors that limit transmission, for purposes of predicting infec- two southern species tend to occur in more permanent water tion distributions. This will prove valuable for health planners bodies. These features of snail distributions suggest that the in the majority of low-income countries where there is a lack of model for S. haematobium is in fact predicting the distribution detailed empirical survey data. Equally important for national of Bu. Senegalensis. Specifically, the model is predicting habi- health planning, the results will help provide estimates of con- tats suitable for this species, i.e., areas with semi-permanent trol program costs of delivering anthelmintics through the water bodies which arose from periodic drying out due to high school system. This has important implications for the effi- temperatures. Moreover, the rarity of water points in the north cient allocation of scarce health resources.
leads to a concentration of human water contacts with fewer Previously, the sensitivity of helminth transmission to cli- water points available, thus increasing the risk of transmis- mate variation, and the use of models using NOAA-AVHRR data sion. In the south, by contrast, human water contacts are more to predict infection patterns, have been described for S. man- dispersed among numerous water bodies, decreasing the risk soni (Malone et al., 1994; Malone et al., 2001). In Egypt, Malone et al. (1994) used 1-km resolution data to derive mapsof diurnal temperature differences (dT), which indicate sur- Estimates of Treatment Population Size and Program Costs face and sub-surface moisture contained in soil and plant can- The best-fit regression models were then used to generate prob- opy and hence may act as a surrogate for the abundance of the ability maps of infection prevalence greater than 50 percent for snail vector, Biomphalaria alexandrina, whereby wetter and S. haematobium, A. lumbricoides, and T. trichiura (Figure 1).
more suitable habitats for Bi. alexandrina corresponded to These maps indicate that different areas would warrant mass lower dT values. They found that low values of dT are associ- treatment with albendazole than those requiring mass treat- ated with increased snail abundance in wet areas with a slow ment with praziquantel—combined control would not be jus- current flow, and is closely mirrored in the patterns of S. man- soni prevalence. Malone et al. (2001) also used 1-km AVHRR We estimate that 5.8 million school-aged children in 33 of data to produce maps of LST and NDVI to study the distribution 49 districts in Cameroon would warrant mass treatment with of S. mansoni in Ethiopia. They found that annual composite albendazole. Using the model for S. haematobium, we estimate maximum LST values of 20 to 33ЊC and wet season values of 18to 29ЊC defined the distribution of S. mansoni prevalencegreater than 5 percent in Ethiopia, and used these limits to pre-dict infection risk within the country. In an analysis of surveydata from Tanzania (Brooker et al., 2001), we have used 8-kmAVHRR to develop predictive models of S. haematobium. Wefound that the model allows reasonable discrimination be-tween high- and low-prevalence schools, at least within thosegeographical areas in which they were originally developed,and performs reasonably well in other coastal areas, but per-forms poorly in comparison in the Great Lakes area of Tanzania.
Despite these applications for S. mansoni and S. haematob-ium, we believe that there are no published studies using satel-lite sensor data to predict distributions of intestinal nematodesin Africa.
Figure 1. Predicted probability of having infection preva- Although the present analysis uses the example of Camer- lence greater than 50 percent. (a) S. haematobium. (b) A. oon because of the geographically detailed data available for lumbricoides. (c) T. trichiura.
the country, the approach can be extended to other countries inAfrica. The potential of such an approach will, however, re- PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING main undefined until further studies are undertaken which Appleton, C.C., and E. Gouws, 1996. The distribution of common intestinal nematodes along an altitudinal transect in Kwa-Zulu consider several issues. The first issue relates to the problem of Natal, South Africa, Annals of Tropical Medicine and Parasitol- spatial scale (Wiens, 1989; Levin, 1992; Walsh et al., 1999). The problem is that many biological responses are scale-dependent(Wiens, 1989), and observed associations between disease Brown, D.S., 1994. Freshwater Snails of Africa and Their Importance, Taylor and Francis, London, England, 608 p.
transmission and environmental variables vary as the scalechanges. In the context of helminth control, the rationale for Brooker, S., M. Booth, and H. Guyatt, 1999. Comparisons of schisto- prediction is to provide information on the spatial patterns of some and geohelminth infection prevalence in school-aged chil- infection and disease at the administrative level at which con- dren from selected areas of Africa: Implications for rapid assess-ment and combined control, Transactions of the Royal Society of trol resources are likely to be mobilized, usually the district Tropical Medicine and Hygiene, 93:125–126.
level. Furthermore, although RS data are available at fine spatialscales, the satellite systems most widely used in the RS com- Brooker, S., and E. Michael, 2000. The potential of geographical infor- mation systems and remote sensing in the epidemiology and con- munity are those with a broad spatial scale of 1 to 8 km. We used trol of human helminth infections, Advances in Parasitology, AVHRR 8-km satellite sensor data to model the probability of areas having infection of 50 percent or greater and warrantingmass treatment with anthelmintics. Other studies conducted Brooker, S., M. Rowlands, L. Haller, L. Savioli, and D.A.P. Bundy, 2000a. Towards an atlas of human helminth infection in sub- at different scales may reach different conclusions (Walsh et Saharan Africa: The use of geographical information systems (GIS), Parasitology Today, 16:303–307.
A further issue is that different environmental variables may impact upon helminth transmission in different areas Brooker, S., C.A. Donnnelly, and H. Guyatt, 2000b. Estimating the number of helminthic infections in the Republic of Cameroon (Appleton and Gouws, 1996; Brooker and Michael, 2000) and, from data on infection prevalence in schoolchildren, Bulletin of in the case of schistosomiasis, different snail species may be the World Health Organization, 78:1456–1465.
differently affected by environmental variability responsiblefor disease transmission (Malone et al., 2001; Brooker et al., Brooker, S., S.I. Hay, W. Issae, A. Hall, C.M. Kihamia, N.J.S. Lwambo, W. Wint, D.J. Rogers, and D.A.P. Bundy, 2001. Predicting the 2002). The development of separate or modified models of spa- distribution of urinary schistosomiasis in Tanzania using satellite tial distribution of infection will provide the basis for a wider sensor data, Tropical Medicine and International Health, and more detailed analysis of the population size at risk of infection and allow for the more targeted and rational imple- Brooker, S., S.I. Hay, and D.A.P. Bundy, 2002. Tools from ecology to mentation of control programs in Africa. The challenge lies evaluate infection risk models, Trends in Parasitology, in press.
however in defining the spatial envelope in which developedmodels can be applied and where different models are re- Bundy, D.A.P., 1997. This wormy world—then and now, Parasitology quired. This is an area of ongoing research.
In summary, the models developed here provide health Bundy, D.A.P., S.K. Chandiwana, M.M.A. Homeida, S. Yoon, and K.E.
planners with a means of predicting the geographical distribu- Mott, 1991. The epidemiological implications of a multiple-infec- tion of intestinal nematodes and schistosomes for the purposes tion approach to the control of human helminth infections, Trans-actions of the Royal Society of Tropical Medicine and Hygiene, of targeting control and for estimating likely program costs.
With the development of further models, the approach couldprovide, for the whole of Africa, reasonable predictions of the Chan, M.-S., G.F. Medley, D. Jamison, and D.A.P. Bundy, 1994. The need to mass treat with albendazole and praziquantel, and evaluation of potential global morbidity attributable to intestinalnematode infections, Parasitology, 109:373–387.
identify areas where a single or combined control approach iswarranted. This will provide a potentially valuable planning Corbett, J.D., and R.F. O’Brien, 1997. The Spatial Characterization tool for planning and implementing control programs aimed at Tool-Africa v1.0, Texas Agricultural Experiment Station, Texas reducing the disease burden due to helminth infections among A & M University, Blackland Research Center Report No. 97-03,Documentation and CD ROM.
Cracknell, A.P., 1997. The Advanced Very High Resolution Radiometer, We thank Andrew Roddam, David Rogers, and William Wint for Crompton, D.W.T., 1994. Ascaris lumbricoides, Parasitic and Infectious providing statistical and GIS advice. We are also grateful to Don Diseases: Epidemiology and Ecology (M.E. Scott and G. Smith, Bundy for comments on the manuscript. The original survey editors), Academic Press, New York and London, pp. 175–196.
work in Cameroon was funded by USAID. S. Brooker and S.I.
Deichmann, U., 1996. African Population Database Documentation, Hay are supported by a Wellcome Trust Prize Fellowship National Center for Geographic Information and Analysis, Santa (#062692) and Wellcome Trust Advanced Training Fellowship Barbara, California, URL: http://grid2.cr.usgs.gov/globalpop/ (#056642), respectively. Data used in this study include data produced through funding from the Earth Observing System EROS Data Center, 1996. GTOPO30 Documentation, Universal Pathfinder Program of NASA’s Mission to Planet Earth in cooper- Resource Locator, Global Land Information System, EROS Data ation with the National Oceanic and Atmospheric Administra- Center, Sioux Falls, South Dakota, URL: http://edcwww.cr.usgs.
tion. The data were provided by the Earth Observing System Data and Information System (EOSDIS), Distributed Active Greer, G.J., R. Mimpfoundi, E.A. Malek, A. Joky, E. Ngonseu, and R.C.
Archive Center (DAAC) at the Goddard Space Flight Center, Ratard, 1990. Human schistosomiasis in Cameroon II. Distribution which archives, manages, and distributes this dataset.
of the snail hosts, American Journal of Tropical Medicine andHygiene, 6:573–580.
Hay, S.I., J.A. Omumbo, M.H. Craig, and R.W. Snow, 2000. Earth obser- vation, geographic information systems and Plasmodium falci- Anderson, R.M., and R.M. May, 1991. Infectious Diseases of Humans: parum malaria in sub-Saharan Africa, Advances in Parasitiol- Dynamics and Control, Oxford University Press, Oxford, England, Hay, S.I., and J.J. Lennon, 1999. Deriving meteorological variables Appleton, C.C., 1978. Review of literature on abiotic factors influencing across Africa for the study and control of vector-borne disease: the distribution and life-cycles of Bilharziasis intermediate host A comparison of remote sensing and spatial interpolation of cli- snails, Malacological Review, 11:1–25.
mate, Tropical Medicine and International Health, 4:58–71.
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING Holben, B.N., 1986. Characteristics of maximum-value composite Ratard, R.C., L.E. Kouemeni, M.M.E. Bessala, C.N. Ndamkou, M.T.
images from AVHRR data, International Journal of Remote Sens- Sama, and B.L. Cline, 1991. Ascariasis and trichuriasis in Camer- oon, Transactions of the Royal Society of Tropical Medicine andHygiene, 85:84–88.
Levin, S.A., 1992. The problem of pattern and scale in ecology, Ecol- Red Urine Study Group, 1995. Identification of High Risk Communities for Schistosomiasis in Africa: A Multi-Country Study, Social and Lindsay, S.W., and C.J. Thomas, 2000. Mapping and estimating the Economic Research Project Reports No. 15, World Health Organi- population at risk from lymphatic filariasis in Africa, Transac- tions of the Royal Society of Tropical Medicine and Hygiene,94:37–44.
Rogers, D.J., 2000. Satellites, space, time and the African trypanosomia- ses, Advances in Parasitiology, 47:130–165.
Malone, J.B., M.S. Abdel Rahman, M.M. El Bahy, O.K. Huh, M. Shafik, and M. Bavia, 1997. Geographic information systems and the Savioli, L., E. Renganathan, A. Montresor, A. Davis, and K. Behbehani, distribution of Schistosoma mansoni in the Nile Delta, Parasitol- 1997. Control of schsitosomiasis—A global picture, Parasitology Malone, J.B., J.M. Yilma, J.C. McCarroll, B. Erko, S. Mukaratirwa, and Stephenson, L.S., 1987. Impact of Helminth Infection in Human Nutri- Xinyu Zhou., 2001. Satellite climatology and the environmental tion, Taylor and Francis, New York, N.Y., 233 p.
risk of Schistosoma mansoni in Ethiopia and east Africa, Acta Stoll, N.R., 1947. This Wormy World, Journal of Parasitology, 33:1–18.
Tucker, C.J., 1979. Red and photographic infrared linear combinations Morgenstern, H., 1998. Ecologic studies, Modern Epidemiology (K.J.
for monitoring vegetation, Remote Sensing of Environment, 5: Rothman and S. Greenland, editors), Lippen-Raven, Philadelphia, United States Census Bureau, 2001. International Data Base, URL: Olds, G.R., C. King, J. Hewlett, R. Olveda, G. Wu, J. Ouma, P. Peters, S. McGarvey, O. Odhiambo, D. Koech, C.Y. Liu, G. Aligui, G.
Venables, W.N., and B.D. Ripley, 1999. Modern Applied Statistics with Gachihi, Y. Kombe, I. Parraga, B. Ramirez, C. Whalen, R.J. Horton, S-Plus, Springer-Verlag, New York, N.Y., 501 p.
and P. Reeve, 1999. Double-blind placebo-controlled study of Walsh, S.J., T.P. Evans, W.F. Welsh, B. Entwisle, and R.R. Rindfuss, concurrent administration of albendazole and praziquantel in 1999. Scale-dependent relationships between population and schoolchildren with schistosomiasis and geohelminths, Journal environment in northeastern Thailand, Photogrammetric Engi- of Infectious Diseases, 179:996–1003.
neering & Remote Sensing, 65:97–105.
Partnership for Child Development, 1999a. The cost of large-scale Watkins, W.E., and E. Pollitt, 1997. “Stupidity or Worms”: Do intestinal school health programmes which deliver anthelmintics in Ghana worms impair mental performance?, Psychological Bulletin, and Tanzania, Acta Tropica, 73:183–204.
———, 1999b. Self-diagnosis as a possible basis for treating urinary Wiens, J.A., 1989. Spatial scaling in ecology, Functional Ecology, schistosomiasis: A study of schoolchildren in a rural area of the United Republic of Tanzania, Bulletin of the World Health Organi-zation, 77:477–483.
WHO, 1995a. WHO Model Prescribing Information. Drugs Used in Parasitic Diseases, Second Edition, World Health Organization, Price, J.C., 1984. Land surface temperature measurements from the split window channels of the NOAA 7 Advanced Very High Resolu-tion Radiometer, Journal of Geophysical Research, 89:7231–7237.
———, 1995b. Health of School Children. Treatment of Intestinal Hel- minths and Schistosomiasis, WHO/SCHISTO/95.112; WHO/ Ratard, R.C., L.E. Kouemeni, M.M. Bessala, C.N. Ndamkou, G.J. Greer, CDS/95.1, World Health Organization, Geneva, Switzerland, 75 p.
J. Spilsbury, and B.L. Cline, 1990. Human schistosomiasis in Cam-eroon. I. Distribution of schistosomiasis, American Journal of Trop- Wright, C.A., 1959. A note on the distribution of Bulinus senegalensis, ical Medicine and Hygiene, 42:561–572.
West African Medical Journal, 8:142–148.
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING

Source: http://seeg.zoo.ox.ac.uk/files/Publications/UsingNOAA-AVHRR_2002.pdf