A comparative study of the prevalence of and factors associated with insecticide-treated nets usage among children under 5 years of age in households that already own nets in Malawi

Background Despite malaria control programmes having successfully increased the number of households owning insecticide-treated nets (ITNs) in Malawi, the population of people with ITN access but still not using them fluctuated from 13% in 2010, 5% in 2012 and then 12% in 2015. This study aimed to compare the rate and factors associated with ITN usage among children under 5 years of age, living in household with at least one ITN, in Malawi between 2010 and 2015. Methods The 2010 and 2015–2016 Malawi Demographic and Health Surveys (MDHSs) were utilized. Only children from households that owned at least one ITN were selected. Multivariate logistic regression analyses were performed to examine associations of child, maternal and household factors with ITN usage. Results In total, 12,378 and 10,196 children under 5 years of age were examined from 2010 and 2015–2016, respectively. ITN usage increased from 57.8% (95% Confidence interval (CI): 56.1%–59.4%) in 2010 to 69.0% (95% CI 67.4%–70.5%) in 2015. The multivariate analysis revealed that, among others, being aged ≥ 24 months, having mothers with no formal education or with primary education, residing in a female-headed household, and residing in households that had poor household ITN supply were significantly associated with reduced odds of ITN usage. Conclusions ITNs are a key vector control intervention in malaria prevention. This study revealed increased ITN usage among children under 5 years old in the 5-year period, suggesting that considerable improvements have been made. However, continued efforts to increase awareness of the importance of using ITNs in malaria prevention in Malawi are necessary. Findings from this research provide some policy implications, especially for improving household ITN supply, to improve ITN utilization in Malawi.


Background
Malaria is one of the leading causes of death in sub-Saharan Africa (SSA) [1]. The disease is caused by the Plasmodium parasite and is transmitted by the female Anopheles mosquito [2]. One of the most important efforts in malaria eradication is the vector control approach which aims at interrupting malaria transmission [1]. Indoor residual spraying with an insecticide to kill mosquitoes, and the use of insecticide-treated nets (ITNs) to prevent mosquito bites are recommended for most of SSA countries. In particular, ITNs are a costeffective intervention [3][4][5][6], and evidence suggests that they are responsible for a 50% reduction in the malaria incidence in malaria-endemic countries [6]. Despite increased efforts to improve ITN accessibility, approximately 437,000 children still die from malaria in Africa every year [7]. The World Health Organization states that the African region is the hardest hit by malaria. For instance, 92 and 90% of all malaria cases and deaths in 2016 worldwide occurred in this region [1]. Additionally, most of these deaths occurred in children under the age of 5 years [1].
Malaria is endemic in Malawi, and there is a huge burden of the disease among children under 5 years old [8]. Malawi experiences perennial malaria transmission with peaks observed in the rainy season between November and April [9]. In 2009, 34% of all outpatient visits, and 40% of all hospitalizations for children under 5 years of age were due to malaria [10,11]. In an attempt to control malaria transmission, in 2001, Malawi was the first country in SSA region to initiate a nationwide ITN social marketing programme [11]. The programme successfully increased ITN ownership and access from 27% and 19% for ownership and access in 2004 to 57% and 38% for ownership and access in 2010, respectively [12]. According to the Roll Back Malaria Monitoring and Evaluation Reference Group, ITN access refers to the proportion of the household population that could have slept under an ITN, assuming each ITN is used by two people and ITN ownership is defined as the proportion of households with at least one ITN [13]. Yet despite the increase in ITN ownership and access, the malaria incidence did not significantly decrease among children [10]. The slow decline in malaria incidence might be attributed to low usage of ITNs among those owning ITNs [12]. Further, a nationwide ITN distribution campaign was conducted in 2012 which increased household ownership of at least one ITN from 58% in 2010 to 70% in 2014 [14].  [12,15]. The surveys used a stratified, two-stage, cluster sampling method to select nationally representative samples. The first stage involved selection of enumeration areas (EAs), as defined by the MPHC. The second stage involved selection of households in each EA using probability systematic selection. In 2010, totals of 20 and 35 households from rural and urban clusters were selected, respectively, while 30 and 33 households from urban and rural clusters, respectively, were selected in 2015.
Data were collected through face-to-face interviews from June to November 2010 for the 2010 survey and from October 2015 to February 2016 for the 2015-2016 survey [12,15]. The survey questionnaires collected information on the mother's demographics, children's illnesses, ITN use, etc. from women aged 15-49 years [13]. The current analysis was restricted to children under 5 years old from households that owned at least one ITN. In this study, data from both the household and mother questionnaires were used.

Outcome variable
The outcome variable was ITN usage defined as whether a child under 5 years of age, living in a household that owned at least one ITN, had slept in an ITN the previous night before the survey. This was assessed as a categorical variable (yes/no).

Independent variables
Three sets of independent variables were considered in the analysis, namely child factors, maternal characteristics and household factors. Child factors included age in months (≤ 12, 13-23, ≥ 24), and gender of the child (male and female). Maternal characteristics included age in years (15-24, 25-34, ≥ 35), educational level (no formal education, primary, secondary/postsecondary), employed (yes, no), ethnicity (Chewa, Tumbuka, Lomwe, Yao, Ngoni, others), marital status (married, unmarried), religion (Catholic, Protestant, Muslim, others), women's health behaviours (undesirable vs desirable). Women's health behaviour was measured using two constructs; number of antenatal care visits (< 4 vs ≥ 4) [16] and baby postnatal check within 2 months (no vs yes). The composite score from the two dimensions ranged from 0 to 2, with 0 (none of the two services) and 1 (one of the two services) representing undesirable health behaviour, and 2 (those who confirmed to have used the two services) representing desirable health behaviours. Finally, household factors included the gender of the head of the household (male, female), roofing materials (improved: cement and roofing shingles; unimproved: no roof, thatch/palm leaf, sod, rudimentary, rustic mat, palm/ bamboo, wood planks, cardboard, wood, others) [17], region (northern, central, southern), and residence (urban, rural). A principal component analysis (PCA) model was used to measure household wealth from a set of materials owned by a household such as bicycles, televisions, etc. Scores were then grouped into quintiles with the upper 20% categorized as richest, the next 20% as richer, middle 20% as middle, penultimate 20% as poorer, and lower 20% as poorest. Household ITN supply was defined as ITN: member ratio within the household [13]. This variable was categorized as very poor (where more than 3 household members had access to 1 ITN), poor (where exactly 3 household members had access to 1 ITN) and good (where exactly 2 household members had access to 1 ITN or where the household had more ITNs than the number of household members).

Statistical analysis
Pooled data from the two survey waves were first analysed to compare differences in the distributions of socio-economic factors between the two surveys using a Chi squared (χ 2 ) test. Thereafter, the survey waves were separately analysed. Further, χ 2 tests were used to examine the relationship between socio-economic factors and ITN utilization. Associations of child, maternal, and household characteristics with ITN non-use were examined using a multivariate logistic regression. Results from the multivariate logistic regression were reported as the adjusted odds ratio (aOR) and 95% confidence interval (CI). The level of statistical significance was set at 5%. All analyses took into consideration the complex sample design of the MDHS by applying the "svy" command. The svy command accounts for sampling weights and clustering effects in the dataset. All analyses were performed using Stata vers. 15.0 (Stata Corp., College Station, TX, USA).

Ethical considerations
The procedures and questionnaires for the standard demographic health survey (DHS) surveys were reviewed and approved by the International Classification of Functioning Disability and Health Institutional Review Board, and ethical clearance was obtained from the Malawi Health Sciences Research. Informed consent was obtained at the start of each interview. Permission to use of the data was sought from the DHS programme.

Distribution of participants by survey year
Totals of 12,378 and 10,196 children under 5 years old were analysed in 2010 and 2015-2016, respectively. Differences between the 2010 and 2015-2016 surveys were significant in terms of the ages of the children, maternal ages, educational levels, employed, marital statuses, health behaviour, gender of the head of the household, and roofing materials (all p < 0.05) ( Table 1). There were no significant differences between the remaining factors between the two survey years.

Distribution of participants according to ITN usage
The rate for ITN use was 57.8% (95% Confidence interval (CI) 56.1%-59.4%) in 2010 compared to 69.0% (95% CI 67.4%-70.5%) in 2015 (Table 2). Table 2 further reveals that in both survey years, ITN usage was associated with a child being ≤ 12 months old, the mother being aged ≤ 24 years, with secondary or post-secondary education, who was married, with desirable health behaviours, and who came from the richest household, living in a male headed household, in urban residence, and with a good ITN supply (p < 0.05). Significant (p < 0.05) differences were also observed between ITN users and non-users in terms of ethnicity in both survey waves. However, having employed mother was associated with ITN usage in 2015-2016 only, while belonging to catholic religious group was associated with ITN usage in 2010 but not in 2015-2016.
In contrast, in 2015-2016, ethnicity, marital status, wealth, roofing material, region, and health promotion did not exhibit significant associations with ITN usage. In 2015-2016, children whose mothers had desirable health behaviours were more likely to sleep under an ITN (aOR = 1.2, 95% CI 1.1-1.3) compared to those whose mothers did not have desirable health behaviours. Additionally, in 2015-2016, belonging to the Protestant religious group and residing in rural areas were associated with reduced likelihood of ITN usage (aOR = 0.8, 95% CI 0.6-0.9) and (aOR = 0.6, 95% CI 0.5-0.8) compared to Catholics and urban dwellers, respectively (Table 3).

Discussion
This is the first nationally representative study to compare ITN usage in children under 5 years old living in households that owned at least one ITN between 2010 and 2015 following the nationwide ITN distribution campaign in 2012. Despite similar ITN access rates between  ITN insecticide-treated net the two surveys (38% for 2010 and 39% for 2015) [12], the rate of ITN usage in children under 5 years old increased from 57.8% in 2010 to 69.0% in 2015. Moreover, the study revealed that many child, maternal and household factors were significantly associated with ITN usage across the two surveys. Specifically, children whose mothers had desirable health behaviours, from households with a good ITN supply, and living in the southern region had increased odds of using ITNs. The increase in ITN usage observed in this study could probably be attributed to two reasons; First, the nationwide ITN distribution campaign in 2012, and a partial distribution campaign in early 2015, which were accompanied by awareness messages on the importance of using malaria interventions such as ITNs. In 2015, a higher proportion (32%) of the interviewed households reported that they sourced their ITNs from mass campaigns than any other sources [12]. Second, the differences in survey periods (i.e. the 2015 survey was largely conducted during malaria transmission peak period (October-February) while most part of the 2010 survey was conducted outside the transmission peak period (June-November). Despite the observed increase in ITN usage, the Malawian ITN usage rate remains relatively lower than that in other countries in SSA such as Rwanda (75%) [18]. The observed difference in ITN usage rates may be as a result of differences in ITN access rates at the time of the study with Rwanda reporting a relatively high ITN access of 64.2% compared to the 39% for Malawi [19]. Additionally, the rates of households owning at least one ITN in Malawi have been fluctuating from 57% in 2010, 70% in 2014, and 57% in 2015 with high rates of ownership observed in the years soon after mass ITN campaigns [12]. The results suggest the need for continued efforts in increasing household ITN ownership and access, through strengthened mass ITN distribution campaigns and routine distribution systems to vulnerable groups.
Children whose mothers were married, had desirable health behaviours, those from middle, rich and richest households, and from households with unimproved roofing were more likely to sleep in an ITN in 2010. Unmarried women may lack support from their spouses and may be disadvantaged in having access to resources [20]. Previous research assessing the association between socio-economic status and net usage revealed conflicting results [5,[21][22][23]. Mothers from rich and middlewealth households may have better access to information and resources that may help them reinforce the decision to have their children sleep under an ITN [24]. Consistent with a previous study in Ethiopia, having improved roofing in a household may foster a false sense of protection from mosquitoes, and hence families may be more likely to forego nets [25]. Women already practicing good health behaviours might be more receptive to other health-promotion practices, such as the use of ITNs [26]. Ethnicity and geographical regions were highly correlated with cultural beliefs and social norms which may influence health behaviours [27]. A higher malaria prevalence was reported in the southern region than the northern part of Malawi [28]; hence, people in the south may feel vulnerable to malaria, thereby making greater efforts to use intervention strategies already in place [28].
The diminishing influence of these aforementioned factors could be related to the mass ITN distribution campaign that was conducted in 2012, which also included a comprehensive awareness campaign. It further appears that the campaign achieved some moderate success in alleviating the influence of socio-economic factors on net usage, as well as in minimizing cultural and regional differences in net usage. This underscores the importance of mass campaigns in reducing disparities in ITN usage behaviour among various segments of the population in Malawi. Future mass campaigns should, therefore, not only emphasize ITN distribution, but also strengthen behavioural change communication messages related to ITN usage.
Additionally, data indicated that older children (≥ 24 months) were less likely to sleep in an ITN compared to those aged ≤ 12 months. It was shown that as children grow older, their access to ITN is lost to younger siblings [29]. Moreover, younger children, especially those that are breastfeeding, are more likely to share a bed with their parents, thereby increasing their chances of sleeping under an ITN [21]. Hence, there is a need to holistically enhance ITN usage among all age groups of children to achieve the malaria elimination goal in Malawi. Mother's education remained a significant factor, consistent with previous research [30]. A mother's level of education may portray the level of autonomy the mother has on important issues such as health decisions [31]. Autonomous women were associated with improved health behaviours [32,33] which may subsequently affect the behaviour of their children [34].
Children from households headed by a female were associated with reduced odds of ITN usage. Households headed by a female may have limited access to resources [35], such as radios, newspapers and television to expose them to important health messages compared to households headed by a male, which may ultimately play a role in their health behaviours, including ITN usage for their younger ones. In addition, a male's position on ITNs might influence household ITN usage as reported in Liberia [36]. Similarly, rural residents might have poor access to health-related information and other resources [3,25,37], hence this explains the finding that children from rural areas being less likely to use ITN compared to urban residents. Moreover, households in rural areas might also face problems of limited public health resources and administrative support within the community. ITN usage is largely dependent on access [13]. This study demonstrated increased odds of ITN usage among children from households with good ITN supply compared to those with very poor ITN supply. Compared to smaller households, larger households face challenges in achieving desirable intra-household ITN supply [38]. This finding underscores the need for malaria programmes to take into consideration family sizes during ITN distribution which may help increase intra-household ITN supply and, ultimately, help improve usage.
In 2015, those belonging to the Protestant religious group had reduced odds of using ITN compared to belonging to the Catholic religious group. This is consistent with a previous study in Ghana where Catholics were more likely to use ITNs compared to those who practiced traditionalism [39]. Efforts to enhance ITN utilization should consider partnership with community religious institutions as an essential approach to scaling up ITN utilization by children.
There are several limitations to this study that should be noted. First, because of the limited number of variables in the MDHS, an array of other important factors related to ITN usage, particularly user-side factors, such as satisfaction with the nets or difficulties associated with using the nets, were not examined. However, this study captured many important child, maternal and household characteristics that could prove vital for effective ITN programmes. Second, the cross-sectional design limited the study from drawing causal inferences between explanatory factors and ITN usage. Third, the study used wealth quintiles which were calculated using PCA model and not multiple component analysis which has been shown to provide more accurate wealth index because of its ability to accommodate both continuous and dichotomous variables [40]. The inability of PCA analysis to accommodate continuous variables makes it difficult to determine wealth quintile cut-off points where most households have the same assets which may result in allocating the same or very similar wealth scores [40]. Finally, the question on ITN usage by children was asked from the mother which could be subject to social desirability bias as mothers may want to prove that they are taking good care of their children by making them sleep under ITNs. Despite these limitations, the study provides evidence of the importance of considering child, maternal and household characteristics, in both research and actual practice, to promote the use of ITNs.

Conclusions
This study examined associations between ITN usage and a comprehensive list of child, maternal and household characteristics. Results demonstrated a considerable increase in the rates of ITN usage among children