Child bed net use before, during, and after a bed net distribution campaign in Bo, Sierra Leone

Background: This analysis examined how the proportion of children less than 5-years-old who slept under a bed net the previous night changed during and after a national long-lasting insecticidal net (LLIN) distribution campaign in Sierra Leone in November-December 2010. Methods: A citywide cross-sectional study in 2010-2011 interviewed the caregivers of more than 3000 under-five children from across urban Bo, Sierra Leone. Chi squared tests were used to assess change in use rates over time, and multivariate regression models were used to examine the factors associated with bed net use. Results: Reported rates of last-night bed net use changed from 38.7 % (504/1304) in the months before the LLIN campaign to 21.8 % (78/357) during the week of the campaign to 75.3 % (1045/1387) in the months after the national campaign. The bed net use rate significantly increased (p < 0.01) from before the campaign to after the universal LLIN distribution campaign in all demographic, socioeconomic, and health behaviour groups, even though reported use during the campaign dropped significantly. Conclusion: Future malaria prevention efforts will need to promote consistent use of LLINs and address any remaining disparities in insecticide-treated bed net (ITN) use.

In 2010–2011, the city of Bo, Sierra Leone, was home to 68 recognized municipal districts locally called ‘sections’. A neighbourhood is officially recognized as a ‘section’ by the municipal government once it reaches a sufficient population size and the community organizes to request formal incorporation. These 68 sections have a footprint of 30.1 km2 (11.6 mi2), and Bo has a relatively high population density for a small African city [21]. Residents within a particular section tend to share not just similar economic and occupational statuses and housing conditions, but also other characteristics such as language, tribe, and religion. After piloting the survey instrument in the two sections nearest to the MHRL research facility, a cluster random sampling method was used to select an additional 18 sections from the remaining 66 city sections across Bo. Population data from the national census bureau suggested that 20 sections would provide more than sufficient numbers of participants to have the statistical power required for the analyses of maternal, child, and environmental health that were the primary goals of the research project. A participatory geographic information system (PGIS) approach was used to create a detailed map of each of the 20 sampled sections, including identifying the location of each of the 1986 residential structures located within these 20 sections [21]. All of these residences were visited by a member of the research team, and adults from each household were invited to participate in the survey. Although most of the 1986 residential buildings in the sampled sections were ‘single-unit’ dwellings (not blocks of flats), the 20 sections were home to 4322 households. Adults living in the homes of their children, siblings, parents, or other relatives were often considered to be members of separate households, especially if they cooked for their own family unit separately from the other residents of the building. A consenting adult from each household within participating residences was asked to provide basic demographic and socioeconomic information about the household, and a supplemental questionnaire was used to collect basic health information about each child in the household who was younger than five years old. In total, 4306 of the 4322 (99.6 %) households in the 20 sampled sections participated in the MHRL health census. These participating households were home to 25,977 individuals. A child questionnaire was completed for 3171 of the 3196 (99.2 %) of under-five children identified as living in participating households. Door-to-door interviews were conducted in the first two sections in Bo during a pilot study from 10 to 24 April 2010. Interviews were conducted in the remaining, randomly-sampled, 18 sections between 1 November 2010 and 11 February 2011. A random number generator was used to select the order in which these 18 sections were interviewed. Interviewing within each section was conducted on consecutive days until all households within the section were contacted, which in some sections meant that some data were collected in more than one of the before-, during-, and after-the-LLIN-campaign time periods. Thus, by chance, data were collected from seven sections before the LLIN distribution campaign, four sections during the campaign, and 13 sections after the campaign. The interviewers—MHRL staff and master’s students studying public health at a local university—all were residents of Bo and were fluent in English as well as a variety of local languages. All interviewers completed a three-day training workshop prior to beginning their fieldwork. Besides providing practical skills in interviewing techniques, data recording, and the use of handheld GPS devices, the training emphasized confidentiality, the informed consent process, respect, and other aspects of research ethics. The pilot study and the expanded study were approved by the research ethics committees of Njala University (Bo, Sierra Leone), George Mason University (Fairfax, Virginia, USA), and the U.S. Naval Research Laboratory (Washington, DC, USA). Participation was completely voluntary, and no incentives or compensation was offered to volunteers. All analyses were conducted using SPSS version 21. Of the 3171 participating under-five children, 123 (3.9 %) were missing responses for the question ‘Did this child sleep under a mosquito net last night?’ and were excluded from the analysis in this paper. Thus, the final sample size for this analysis was 3048 under-five children. Two-sided Pearson Chi squared tests with a significance level of α = 0.05 were used to examine possible differences in reported bed net use before and after the LLIN campaign within various population subgroups, such as comparing before and after rates among 2-year-olds or among those without electricity in the home. Logistic regression models were used to examine the predictors of bed net use separately for the before-, during-, and after-the-campaign time periods. Because there were some covariates for which exposure rates differed significantly by age, such as infants (those less than 1 year old) being significantly less likely than older children to have been vaccinated for measles, dummy variables for age in years were used in all logistic regression models. Dummy variables for the surveyed sections were also included in the regression models in order to adjust for neighbourhood-level differences in socioeconomic status, population density, proximity to the centre of Bo city, and access to health resources that might not adequately be accounted for solely by household-level variables such as housing construction materials and access to utilities. For each demographic, socioeconomic, or health behaviour variable (all shown in Table 1), three different models were fit, one each for before, during, and after the LLIN campaign. Each model included the exposure of interest, age dummies, and section dummies. The p value for each sub-population in its own time-specific model is shown in Table 2. Factors associated with children less than 5 years old sleeping under a bed net before, during, and after the national LLIN distribution campaign in Bo, Sierra Leone Italicized text indicates statistically significant results (p < 0.05) Odds ratios and 95 % confidence intervals for differences in bed net use among those with and without various characteristics, after adjusting for age and section (neighbourhood) Italicized text indicates statistically significant odds ratios

N/A

• ranasinghe, shamika
• ansumana, rashid
• bockarie, alfred s.
• bangura, umaru
• jimmy, david h.
• stenger, david a.
• jacobsen, kathryn h.