Spatial and statistical prediction of urban malaria in Yaoundé [Elektronische Ressource] : a social and environmental modelling approach for health promotion / von Roland Ngom

Ngom M.*

Informations

• Publié par: padagogische_hochschule_heidelberg
• Publié le: 01 janvier 2010
• Nombre de lectures: 68
• Langue: English
• Poids de l'ouvrage: 15 Mo

Extrait

 
Spatial and Statistical Prediction
of Urban Malaria in Yaoundé:
A Social and Environmental Modelling Approach
for Health Promotion
  
 
Von der Pädagogischen Hochschule Heidelberg
zur Erlangung des Grades eines
Doktor der Philosophie (Dr. phil.)
genehmigte Dissertation von


Roland Ngom

Aus
Yaoundé

2010

 
   
 
Erstgutachter:
Prof. Dr. Alexander Siegmund (Pädagogische Hochschule Heidelberg)

Zweitgutachter:
Prof. Dr.med. Thomas Kistemann (Universität Bonn)

Fach:
Geographie und ihre Didaktik

Tag der mündlichen Prüfung:
11. Oktober 2010










Nothing can be more important to a State than its public health; the State’s paramount
concern should be the health of its people.

Franklin D. Roosevelt
(In a report to the new Legislature, 1931)




























Gedruckt mit Unterstützung des Deutschen Akademisches Austauschdienstes
Printed with the kind assistance of the German Academic Exchange Service









To my beloved wife and children who
supported my long absence during the realisation of this study

To my beloved mother and sisters for taking care of me during these moment.

To my family in law in Cameroon

To my brother

To my deceased father for inspiring me
III

Acknowledgements

My childhood in Cameroon, and memories of people suffering from various illnesses,
particularly malaria, were the initial factors that pushed me to be involved in geomedical
studies. My desire to bring my modest contribution to a better public health status on the earth
in general, and in Cameroon particularly, is equivalent to my scientific interest in medical
geography and malaria studies. Sharing knowledge with people has always been a part of my
wishes, which I have tried to express in a useful scientific way in this study.

Thanks to the almighty God. Many thanks to Prof. Dr. Alexander Siegmund for enrolling me
as a scientific assistant. His engagement in this project, his permanent availability and his
valuable advice were decisive elements. Thanks to Prof. Dr. Maurice Tsalefac of the
department of Geography at the University of Yaoundé I for his valuable scientific and
logistical assistance during my stays in Yaoundé. Thanks to Prof. Dr. Kistemann for his
assistance. Many thanks to Jeanne Mbousnoum, teacher in a high school in Yaoundé. She
introduced me into schools in Yaoundé, and gave me useful advice. Thanks to Dr. Alain Biem
of IBM in New York, and Dr. Valerie Louis of the Faculty of Medicine in Heidelberg for their
precious advices during our many scientific discussions. I also thank Prof. Dr. Dieter Hupke
and Prof. Dr. Michel Ulrich for their availability. Thanks to all my colleagues in the
department of Geography for always being ready to assist me in this project. Thanks to the
students of the department of Geography of the University of Yaoundé I, who agreed to be
part of the (GeoMedical Information System) GeoMedIS project. Thanks to all the people
who freely gave their time and intimate private space in order to allow the test of antimalarial
games, and to answer questions during the interviews. Thanks to all the interviewers and
students who helped in managing the numerous questionnaires. Thanks to Annette Wakam for
assisting me during critical periods.

Special thanks to the Deutsche Akademische Austauch Dienst (DAAD) for their generous
financial support. Thanks to the “Vereinigung der Freunde der Pädagogische Hochschule
Heidelberg e.V.” for supporting the financial cost of remote sensing data.
IV

Abstract

Most of the existing predictive malaria risks models use very broad spatial scales. They are
usually built for continental or national outlines. These models do not account for the
complexity of socio-economic variables intervening in the malaria transmission process. Most
of them are driven by weather data. However, it is difficult to make antimalarial interventions
at a continental or national level and to act on climate variables alone. Consequently, the
suitability of these models for real malaria prevention strategies is not high. Moreover, the
existing informational-based prevention strategies are not suitable, since they are usually
limited to the occasional usage of large public mass media to transfer bits of information.

This study proposes new paths in malaria modelling and prevention. It is dedicated to the
building of a thematically extended model integrating both environmental and social
variables. The proposed prevention strategy is based on an educational philosophy integrating
the actual malaria modelling results. The study starts with the implementation of a
methodology dedicated to data creation and data analysis. The protocol of data creation is
based on an urban malaria paradigm. It encompasses the epidemiological, environmental, and
social components of malaria risk. The epidemiological component is elaborated through
retrospective, self-reported, malaria febrile and clinical episodes of individuals at the
household level. In addition to climate data, key ecological variables are created from remote
sensing sources with a very high spatial resolution. New social indexes and coefficients
measuring economic status, crowding conditions and prevention capacity of the population
are created. A morphospatial structure of Yaoundé, which assumes the presence of distinct
population aggregates, representing similar socio-economic profiles, is established using an
object-oriented classification of QuickBird images. A spatial based index of urbanity (IU),
quantitatively marking the difference between “urban” and “rural” patterns, is also built. A
knowledge-base expressing the social, ecological and malarial significance of both population
aggregates and index of urbanity is established and used in a Fuzzy Logic simulation
approach to predict urban malaria in Yaoundé.

The yearly malaria prevalence based on individuals in households in Yaoundé is 9% while the
malaria prevalence based on households as an entity is 27%. Malaria prevalence is higher
during the small rainy season. It is much more marked in peri-urban areas during this season,
while people in central areas are more exposed during the big rainy season. A statistical
multinomial model identified socio-economic and socio-ecologic variables, notably those
related to the physical condition of houses, as being highly associated with frequent episodes
of malaria in households. Variables related to prevention capacity perform very well in
predicting the absence of malaria in households. Among the ecological variables, only
elevation and the distance to urban agriculture (UA) areas are associated with malaria. The
statistically (from multinomial models), overall-predicted household prevalence of malaria is
lower than the observed one. The morphospatial structure of the city shows a clear distinction
between very dense, centralized and “urbanized” population aggregates (PA) and very
isolated, mostly peri-urban, “rural” population aggregates. The morphological model suggests
that the intensity and sustainability of the malaria transmission are both dependant on
demographical gradients. The less urbanized population aggregates, although being in
proximity of urban agriculture areas, are demographically not suitable for a sustainable
malaria transmission. The most urbanized population aggregates are too dense and too far
from urban agriculture areas. This does not favour local malaria transmission. This rigid
ecological pattern is somewhat biased by the identified social patterns. The densest population
aggregates mostly host very poor people. This allows a part of this population to be at a high
V

risk of malaria through localized urban farming activities. The other parts of dense population
aggregates are located in centrally situated planned zones. They have better socio-economic
and socio-ecologic conditions which significantly reduces their vulnerability to malaria.
Population aggregates with suitable demographic conditions (not too dense or too isolated), in
addition to a higher environmental and social vulnerability, are the most exposed. Although
the Fuzzy Logic simulation procedure produces a predicted prevalence which is lower than
that of the overall multinomial model, it also identifies these intermediate population
aggregates as being the most exposed.

Results of interviews show that, in general, the knowledge and perception of people of basic
key factors associated with malaria transmission is bad. Moreover, this knowledge follows a
social stratification with the richest people having the best background. An association
between knowledge and prevention behaviour is also established. In order to use the model
fo