Housing conditions and Plasmodium falciparum infection: protective effect of iron-sheet roofed houses
© Yé et al. 2006
Received: 01 September 2005
Accepted: 01 February 2006
Published: 01 February 2006
Skip to main content
© Yé et al. 2006
Received: 01 September 2005
Accepted: 01 February 2006
Published: 01 February 2006
Identification and better understanding of potential risk factors for malaria are important for targeted and cost-effective health interventions. Housing conditions have been suggested as one of the potential risk factors. This study aims to further investigate this risk factor, and is focused on the effect of the type of roof on Plasmodium falciparum infection among children below five years in the North West of Burkina Faso.
In a cross-sectional study design, 661 children aged six to 60 months were randomly selected from three rural and one semi-urban site at the end of the rainy season (November 2003). The children were screened for fever and tested for Plasmodium falciparum infection. In addition, data on bed net use and house characteristics was collected from the household were each child lived. Using adjusted odds ratios, children living in house roofed with iron-sheet were compared with those in house with mud or grass roof.
Overall P. falciparum infection prevalence was 22.8 % with a significant variation between (Chi-square, p < 0.0001). The prevalence in Cissé (33.3 %) and Goni (30.6 %) were twice times more than in Nouna (15.2 %) and Kodougou (13.2 %). After adjusting for age, sex, use of bed net and housing conditions, children living in houses with mud roofs had significantly higher risk of getting P. falciparum infection compared to those living in iron-sheet roofed houses (Odds Ratio 2.6; 95% Confidence Interval, 1.4–4.7).
These results suggest that house characteristics should be taken into consideration when designing health intervention against P. falciparum infection and particular attention should be paid to children living in houses with mud roofs.
Malaria, a preventable disease, is still one of the most important causes of morbidity and mortality in developing countries, especially in sub Saharan Africa . For decades, a number of initiatives have been undertaken to control the disease and reduce it related burden. Unfortunately, the success of these actions remains restricted to some specific areas [2, 3]. An increase in the malaria burden is envisaged due to drug resistance, breakdown of health care systems and environmental changes . The limited success of a global approach to malaria eradication can be attributed to the difference in epidemiology, environment and socio-economic conditions from one setting to another. There is, therefore, a need to understand context-specific potential risks factors associated with Plasmodium falciparum infection.
Among the list of risk factors, ranging from biological, physical environment, health care systems and socio-economic conditions, the design of a house significantly affects the incidence P. falciparum infection [5–8]. Gamage-Mendis et al  demonstrated that in Sri Lanka, living in a completed house with brick and plaster walls and tiled roofs was highly associated with malaria risk reduction compared to living in the poorest type of house. In a randomized controlled trial in rural Gambian, Lindsay et al  found that adding ceilings in mud huts reduced the number of Anopheles mosquitoes and other vectors entering the room and may be an effective way to reduce malaria risk.
Better housing is one of the factors that reduced malaria infection risk in regions that used to be endemic and modification of houses was used to protect against malaria in Italy, Greece, Panama and the USA in the early 20th century . Alongside life style and local habits, housing conditions play a role in modulating exposure of populations to mosquitoes and vector-borne diseases .
Roof types determine indoor temperature, a parameter that is associated with malaria transmission. The objective of this study was to investigate the effect of the type of roof on P. falciparum among children below five years of age in the North West of Burkina Faso. This was done because, considering current knowledge, no study has specifically assessed and quantified the effect of different types of roofs upon P. falciparum infection in a sub-Saharan setting.
All the sites were part of a Demographic Surveillance System (DSS), which covers the Nouna health district located in the province of Kossi. In this area, since 1992 the population is followed longitudinally for demographical events (birth, death, in and out-migration) . These data were made available for this study.
Study participants (661) were children aged between six and 60 months from the four sites. They were selected randomly through cluster sampling of households (352), from the four sites, the sampling frame of which was obtained from the Nouna DSS database.
Parents of selected children were visited prior to the survey to obtain informed consent. The objective and methods of the study were explained to them during these visits. They were also told that the finger prick procedure was very safe with the use of single-use lancets, although it may be slightly painful for the children; that the quantity of blood taken with this method was not harmful for the child; that the child would benefit from a medical check-up for malaria and other common pathology; that medical treatment would be offered free of charge if needed; and that referral was possible for severe cases.
Malaria infection data were collected by active case detection from each of the four sites from the 14th to 18th November 2003. A mobile team of one physician, one nurse, two laboratory technicians and five interviewers visited each site. Mothers were asked to bring their children to a meeting point identified by the village chief. Children were screened for fever using digital axillary thermometers. Febrile (body temperature > = 37.5°C) children were physically examined, including weight measurement and spleen palpation and then treated with chloroquine (25 mg/kg weight during three days: day one: 10 mg/kg, single intake; day two: 10 mg/kg, single intake; day three: 5 mg/kg) and paracetamol. A blood sample was then taken using the finger prick method. Severely sick children were referred to the nearest health unit. The slides were later stained with Giemsa and read for parasite count in the laboratory of the "Centre de Recherche en Santé de Nouna" (CRSN). Parasite density was estimated by counting 100 fields and equating this to 0.25 μl of blood . Each child with any P. falciparum detected was classified as infected.
In addition, mothers were asked if the child had been sleeping under a mosquito net, and whether the mosquito net was treated or not with insecticide. Use of anti-malarial drugs in the previous week was also noted.
After the completion of the clinical phase, mothers were visited at home by the interviewers for detailed interviews about their housing conditions. Since a typical household compound in this area is comprised of several houses, the house where the participating child usually slept was selected. Using a structured questionnaire, the interviewer asked about the type of material used for the wall (mud block, grass, stone, or cement bricks) and the roof (iron-sheet, mud or grass). They also asked about the presence of animals, well and potential mosquito breeding site (open water body) within 50-meter radius of the house. All answers were cross-checked by observation.
Proc logistic procedure was used to estimate the odds ratio (OR) of living in a house with a mud roof and one with a grass roof compared to living in a house with iron-sheets. The multivariate model included potential confounders such as age group (categorical: <12, 12–23, 24–60 months), location of the house (categorical: Cissé, Goni, Kodougou, or Nouna), presence of animal (binary: yes or no), presence of a well (binary: yes or no), presence of breeding sites and use of mosquito net (binary Yes or No). The logistic regression model was defined as follows:
log it (π i ) = β 0 + β 1 Roof_earth i + β 2 Roof_grass i + β 3 X i + β 4 X i +...+ β x X i
Where π i is the predicted probability for the i th child, of being positive Pf infection or having clinical malaria. The odds of the same child will be , β 0 the intercept, and β 1... β x the regression coefficients of the independent variables (name following each coefficient X i being the potential confounders mentioned above). The odds ratio associated with Roof_earth compared to iron-sheet roof (reference) is the exponential of β 1 (OR Roof_earth = exp(β 1)). The significance of the OR was assessed using 95% confidence intervals (CI). All statistical analysis was done using SAS version 9.1 (Cary, NC USA).
Distribution of the study participants, by site, sex and age, November 2003
# of Households
Distribution of participants according to bed net use and housing conditions, November 2003
Reported use of bed net
Animal within 50 m radius of the house
Well within 50 m radius of the house
Breeding site within 50 m radius of the house
Reported mosquitoes net use among participants was relatively high (53.3 %) but varied significantly (chi-square, p < 0.0001) between sites with the highest coverage in Goni (80.4 %) and Kodougou (68.3 %). Nouna had surprisingly the lowest coverage (15.6 %). Data on mosquito net use was missing for 3.8 % of the children and these were eliminated from subsequent analyses.
Overall, 61.9 % of children lived in a house with an animal enclosure within a 50-meter radius but there was significant variation between the sites with a maximum in Cissé of 80.9 % (Chi-square, p < 0.0001). 18.5 % of households had wells and 13.5 % had a potential breeding site within a 50-meter radius of the house. In Nouna these figures were significantly higher than at the others sites 50.6 % and 40.9 % respectively (Table 2).
Distribution of P. falciparum infection prevalence among study participants by sex, age and site, November 2003
Effect of house roof on Plasmodium falciparum infection among children
One of the principal issues in public health is the question of equity of treatment and variation in vulnerability of the relevant population. Of special relevance is the effect of socioeconomic class on the risk of morbidity and mortality. Malaria as a "poor man's disease", which affects either the children or parents in poverty stricken households of sub-Saharan Africa, epitomizes the problem. In this research one characteristic of financial status was tested, the quality of housing. Corrugated iron-sheets are relatively expensive and cannot be afforded by most villagers, in spite of their better protection from dampness. The question was whether this roofing material would also protect against malaria. There were two conflicting hypotheses as to the influence of the roof on the preponderance of house-dwelling mosquitoes, and hence on mosquito prevalence: (a) the iron-sheet would increase prevalence, due to the heating of the roof and resulting in increased indoor temperature and (b) the iron would decrease prevalence as the roof would no longer be a suitable resting place for blood-engorged mosquito, as opposed to cracks in the mud roof, and this would disturb the malaria-transmission cycle.
The finding that there is a significant increase in malaria risk in mud roofed huts, after adjusting for other environmental conditions, suggests that the poorer sections of the population are at greater risk and thus require more intensive prevention. In this case it seems that internal spraying of roof areas would be useful. In addition, in areas where these types of dwellings are common, intensive surveillance would be required to reduce the high levels of child morbidity and mortality. From a biological point of view, it is expected that the microclimate conditions in grass roofed houses are more or less similar to the ones in mud roofed houses and similar precautions should be recommended. The non-significance of the results for the grass roofed houses in this case may be attributable to the small proportion of children living in this type of house, resulting in low statistical power.
At a more theoretical level, weather-based models of malaria incidence, prevalence and mortality, require a clear evaluation of the ambient conditions of the mosquito environment. While it has been suggested that inside-outside temperature differentials are important, the results here suggest that the structural conditions are of greater importance.
Other possible explanations for the differences in prevalence between roof structures may also be raised. For instance, the more affluent houses may have better sealing around doors, hence preventing entry of mosquitoes. In addition, in the more affluent households children may be better nourished and hence less susceptible to infection. On the other hand, socioeconomic status would not have had much impact on the results since the main outcome was P. falciparum infection rather clinical malaria. The latter is likely to be influenced by better access to treatment. Further research will be required to test these explanations.
The paper presents the risk of P. falciparum infection associated with the type of roof among under five children in North West of Burkina Faso, a holoendemic area. P. falciparum infection prevalence is consistently high in the four sites although there are differences between sites. After controlling for potential confounders, it was shown that children living in iron-sheet roofed houses have two times less risk of getting P. falciparum compared to those living in houses with mud roofs. This study adds evidence to the body of knowledge regarding the effect of house type on P. falciparum infection risk. In conclusion, taking house characteristics into consideration in any health intervention programme against malaria will be beneficial.
The authors would like to express their gratitude to the fieldwork team (Dr Florent Somé (RIP), Dr Boubacar Coulibaly, Mohammed Diaby, Dissa Sory Ibrahim, Omer Kiénou, Mêman Sanogo, Jean Marie Somé and Dieudonné Zerbo), to the parents and their children who participated in the study. We would also like to thank Dr. Catherine Kyobutungi who proofread the manuscript. The study was funded by Deutsche Forschungsgemeinschaft (DFG) and Union des Banques Suisses (UBS) Optimus Foundation
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.