Determinants of insecticide-treated net ownership and utilization among pregnant women in Nigeria

Background: Malaria during pregnancy is a major public health problem in Nigeria leading to increase in the risk of maternal mortality, low birth weight and infant mortality. This paper is aimed at highlighting key predictors of the ownership of insecticide treated nets (ITNs) and its use among pregnant women in Nigeria. Methods. A total of 2348 pregnant women were selected by a multi-stage probability sampling technique. Structured interview schedule was used to elicit information on socio-demographic characteristics, ITN ownership, use, knowledge, behaviour and practices. Logistic regression was used to detect predictors of two indicators: ITN ownership, and ITN use in pregnancy among those who owned ITNs. Results: ITN ownership was low; only 28.8% owned ITNs. Key predictors of ITN ownership included women who knew that ITNs prevent malaria (OR = 3.85; p < 0001); and registration at antenatal clinics (OR = 1.34; p = 0.003). The use of ITNs was equally low with only 7.5% of all pregnant women, and 25.7% of all pregnant women who owned ITNs sleeping under a net. The predictors of ITN use in pregnancy among women who owned ITNs (N = 677) identified by logistic regression were: urban residence (OR = 1.87; p = 0.001); knowledge that ITNs prevent malaria (OR = 2.93; p < 0001) and not holding misconceptions about malaria prevention (OR = 1.56; p = 0.036). Educational level was not significantly related to any of the two outcome variables. Although registration at ANC is significantly associated with ownership of a bednet (perhaps through free ITN distribution) this does not translate to significant use of ITNs. Conclusions: ITN use lagged well behind ITN ownership. This seems to suggest that the current mass distribution of ITNs at antenatal facilities and community levels may not necessarily lead to use unless it is accompanied by behaviour change interventions that address the community level perceptions, misconceptions and positively position ITN as an effective prevention device to prevent malaria. © 2011 Ankomah et al; licensee BioMed Central Ltd. The study was a population-based household cross-sectional survey undertaken in October-December 2008. Multi-stage cluster sampling design was used to select pregnant women. The selection process was aimed at achieving a representative sample of study participants at the state level. Structured questionnaire based on thematic areas was pilot-tested before use. The study questionnaires elicited information on demographic socio-economic, knowledge, attitude, belief, health care-seeking behaviour and the various issues related to prevention and treatment of malaria in pregnancy. Interviewers were selected based on their understanding of both English and the local languages of the communities where they worked. Interviewers were trained to understand and appropriately apply all aspects of the questionnaire and interview process. Approval was sought and received from each State Ministry of Health, community leaders, and informed consent was obtained from the heads of households and the respondents. All interviews took place in the homes of respondents and confidentiality was assured. Out of the 36 states that make up Nigeria, 21 (including the Federal Capital Territory) were included in this survey. These 21 states were selected because they had fairly comparable access to resources for state-wide implementation of interventions for malaria control (including distribution of free ITNs to pregnant women). Eighteen of the states were the only states in the country supported by the Global Fund grants at the time of the survey while the other three were receiving comparable support from leading international development partners. In addition to the donor support, all the states received statutory funding from annual government budgets. The Malaria Control Programme in all these states followed a National Strategic Plan that derived largely from the Global Malaria Control Strategy and Roll Back Malaria. Resources provided for malaria control include long lasting insecticide treated bed nets, artemisinin-based combination treatments (ACTs) and intermittent preventive treatment in pregnancy. We used a multi-stage sampling procedure to select 20 clusters per State. To achieve this, we used enumeration area (EA) maps for the Local Government Areas (LGAs) and communities selected from each state. The EA maps were obtained from the National Population Commission. A cumulative measure of the size was obtained using the appropriate power calculation formulae. We estimated that 20 clusters of approximately 18 households per cluster would yield a minimum of 120 pregnant women per state. Women of reproductive age within the selected households were identified and listed. From the household listing, women of reproductive age were identified and further screened to select those who were pregnant and these were subsequently interviewed. The total sample size was 2348 pregnant women. Data were entered using the Census Surveys Professional software (CSPro), and were exported to, managed and analyzed using SPSS 18.0®. We employed a logistic regression to examine the factors that predict two outcomes; first, bednet ownership among the total sample of pregnant women, and second, the use or non-use of ITNs among pregnant women who own bednets in Nigeria [25,26], This permitted us to compute the probability that a respondent used or did not use ITNs controlling for other covariates, including demographics and knowledge indicators. Conceptualization and construct of variables and analytical models took into consideration the key concepts of the "health belief theoretical model" namely perceived susceptibility, severity of perceived threat and perceived benefit of recommended health action which, in the context of this study refers to ownership and use of ITNs during pregnancy (25, 26). The study has two dependent variables: net ownership; and the use of bednets by pregnant women the night before the study. In this study, the definition of ITNs includes long lasting insecticide treated nets and all other insecticide treated nets. Therefore, all insecticide treated nets, whether long lasting insecticide treated nets or retreated nets are referred to in this paper as ITNs. Non treated nets were not included in the definition of use. The independent variables in this paper are: geo-political zones (Nigeria has six geo-political divisions: North-East, North-Central, North-West, South-East, South-South and South-West), residence (urban or rural), age, education, knowledge about cause of malaria (i.e. knowing that it is transmitted by mosquitoes). Composite measure for 'misconceptions about causes of malaria', and 'misconceptions about prevention of malaria' were based on variables identified from previous studies [23,24]. These included the attribution of malaria to 'staying too long in the sun', 'not resting enough/no sleep', 'drinking too much alcohol/beer', 'witchcraft/juju', 'eating too much palm/groundnut oil', and 'excessive laborious work. Knowledge about malaria prevention was created as a composite indicator measuring how pregnant women responded to whether or not malaria could be prevented by taking any of the following actions: taking malaria medicine every day/weekly; Clearing bushes around the house, destroying places where mosquitoes breed (e.g. stagnant water), sleeping under insecticide treated net, using mosquito repellent creams/mosquito coils, keeping out of the sun, using insecticide to kills mosquitoes before sleeping in a room, using nets on windows and doors of the house, and using brooms/rackets. One item - knowledge that ITNs prevent against malaria- was analysed separately. The analysis of the determinants of each of the two dependent variables was done separately. Both bivariate and multivariate analyses were undertaken to explore the determinants of bednet ownership and the statistical significance was based on a p-value of less than 0.05. At an exploratory data analysis stage, several models were explored. Test of goodness-of-fit was based on Hosmer and Lemeshow test (HL); models with p > 0.05 were assumed to be good and fit well to the data. Several models were explored but the model with the best fit according to HL was selected for reporting (p-value = 0.832).