Background

Malaria remains a major cause of morbidity and mortality among children less than five year old particularly in Sub-Saharan Africa despite several decades of exposure to malaria control interventions. In 2012, The World Health Organization (WHO) estimated that 90% of all malaria deaths worldwide occurred in the Africa region and mostly among children under-five years of age (under-fives). Although malaria-specific mortality rates among children in the region have declined by an estimated 54% since 2000, high endemicity persists in many Sub-Sahara African countries including Nigeria [1] where it accounts for about 110 million clinically diagnosed cases per year, an estimated 60% of annual outpatient visits and 30% of all hospitalizations., About 300,000 children in Nigeria die from malaria annually [1]. About 25% of infant mortality and 30% of mortality in children under-five years in Nigeria are attributed to malaria [1].

Although malaria has been eliminated in Europe, Asia and the Americas, efforts to eradicate malaria in Sub-Saharan Africa have largely been less than successful [2]. Regional and international efforts to fight malaria since the 1940s have recently been subsumed into a coordinated global response. In 1998, WHO, UNICEF, United Nations Development Program (UNDP) and the World Bank launched the Roll Back Malaria (RBM) partnership to achieve universal coverage of malaria control interventions. The Millennium Development Goals (MDG) Malaria summit in 2008 endorsed a Global Malaria Action Plan aimed at achieving a substantial reduction in the burden (elimination in 8–10 countries including Nigeria, and the eventual eradication) of malaria [3]. Since its inception in 2002, the Global Fund to Fight AIDS, Tuberculosis and Malaria established as an international financing mechanism to fight malaria (including AIDS and Tuberculosis), have thus far committed over US$17billion in funds to about 144 countries and have supported the distribution of 122 million bed nets and financed the delivery of over 142 million drug treatments encounters to fight malaria. In addition, it has committed over $ 8.33 billion to malaria-related initiatives in 80 countries [4]. The Bill and Melinda Gates Foundation has supported the development of new tools to combat malaria through innovations and scientific research [5].

Malaria has exacted severe socio-economic burden on countries and malaria endemic communities. In Nigeria, an estimated 132 billion Naira (2013 USD 825 million)—in treatment costs, prevention, loss of work time—is lost to malaria annually [6]. Nigeria developed a comprehensive National Malaria Control Strategic Plan (NMCSP) to “reduce deaths and prevalence of malaria in children under-five years old; increase ownership and use of insecticide-treated nets (ITNs) and long-lasting insecticidal nets (LLINs); increase the use of diagnostic tests; and improve appropriate and timely treatment of malaria; among other strategies”[6]. Millions of ITNs and LLIN have so far been distributed in Nigeria. According to the Nigerian Demographic and Health Survey (NDHS) 2013, the percentage of households in Nigeria with at least one of any type of mosquito net rose from 17% in 2008 to 55% in 2013[1]. Similarly, during the same period, households with at least one insecticide-treated net (ITN) rose from 8% to 50%. Household possession of at least one bed net was higher in rural households (61%) than in urban households (48%) [1]. Inspired by the Millennium Development Goals (MDGs), international partners notched up support to national programs in the distribution of ITNs, particularly to rural Nigerian communities.

Despite the development of these proven, preventive interventions to reduce the burden of malaria and the availability of effective treatment technologies, access to effective management of malaria remains poor in Nigeria. Evidence has shown that community case management (CCM) of malaria by appropriately trained, well-supervised community health workers supported with uninterrupted supply of medicines, can help reduce overall and malaria-specific under-five mortality by between 40 and 60 per cent, respectively, and severe malaria morbidity by 53%. [7–9] Unicef, WHO and partners have actively supported integrated community case management strategy in an increasing number of countries to train, supply and supervise front-line providers to treat childhood malaria in malaria-affected countries using artemisinin-based combination therapy (ACT). In addition, community level testing for malaria has been facilitated by the availability of high-quality rapid diagnostic tests (RDTs) for malaria. [9]

Methods

A case controlled, pre- and post-intervention study was design to evaluate the integration of Patent Medicine Vendors (PMVs) into integrated community case management (iCCM) of childhood diseases including malaria. Using a multi-stage stratified cluster design, data was obtained data from 3,910 sampled households: 2277 and 1633 households in the pre- (baseline survey was conducted in November 2013) and post-intervention periods (end-line survey was conducted in January 2014) respectively, across the two control (Alkaleri, Ningi) and two intervention (Tafawa-Balewa, Ganjuwa) local government areas (LGAs). The control LGAs and the intervention LGAs are located in the northwest and southeast regions of the State. They were selected purposively to avoid shared boundaries to minimize spillover effects and for their accessibility to Bauchi City, the capital city. Each. These LGAs are located on the east, west, south and central geo-political regions of the state and were selected for the study for their accessibility from the central administrative region. Random samples of cluster (national census enumeration areas) were obtained from each of the four LGAs. Samples of household were obtained from the clusters through systematic randomization. All the under-five year olds in each household were included in the study. Using a pretested questionnaire administered on the mothers (caregivers), the study captured data on household socio-demographic characteristics of the children and their caregivers including the management of malaria, the number, types and duration of possession of the bed nets. The history of fever was used as a proxy for malaria fever similarly to the NDHS 2013 [1]. Caregivers were asked if any child under five year old in the household had had fever, diarrhea or cough in two weeks preceding the interview. They were then asked where they had sought care and what treatments including medications they were given. Data from 5,814 under-five children from sampled households—3077 and 2737 in the pre-and post-intervention periods—were obtained for the study.

Analysis

Contingency table analysis was used to partition the cases with fever from non-fever cases classified by exposure or non-exposure to bed nets, to estimate and compare the odds ratio of malaria fever in children who lived in households with insecticide treated bed nets with those who lived in households without insecticide treated bed nets. The odds ratios of fever in the pre- and post-intervention periods, in the intervention and control LGAs were also estimated. In addition, fever prevention fraction attributable to bed nets usage, and the interventions were estimated to demonstrate the impact of the interventions. We implemented a logistic regression model (Eq 1), and used fever/no-fever (F_it) as the outcome variable, at study group i (control LGA = 0, intervention LGAs = 1) and period t (pre-intervention = 0, post-intervention = 1) against the study-group (I_it) and period (P_it) dummies including interaction between study-group and period ((P*I)_it). This interaction term is an estimator of the prevention (or attributable) fraction and the overall impact of the interventions on reduction of the prevalence of fever. This was essentially a difference-in-differences analysis to adjust the impact of the interventions for the changes that may have occurred unprovoked in the control group between the pre- and post-intervention periods. The estimates derived from the logistic model improved on the contingency table estimation in that the logistic regression allowed for further stratification of the explanatory variables, and controlled for the confounding effects (C_it).

(1)

The age and educational background of the mothers (caregivers), the number of children under-five years who lived in a household [12, 13] and the socio-economic background [14] of households can influence household demand for, possession and use of bed nets on the one hand [14, 15] and the risk of contracting malaria on the other [15]. For instance, economically advantaged households may have had better access to bed nets and improved housing conditions that reduced the risk of parasite transmission than the less economically advantaged households. We used Principal Component Analysis (PCA) to construct an index of household socioeconomic wealth from household durable assets [16–18]. The wealth index scores were then used to categorize households into to low-, middle- and high-socio-economic status included as an independent variable in the analysis. The logistic regression model was weighted with inverse probabilistic sample weights to account for differential probabilities in each sample because of the nature of the multistage cluster design and to improve the accuracy of the estimates. Analyses were performed with STATA^(C) version 12; statistical level of confidence was set at 95%. Bayesian and Akaike’s Information Criteria (BIC & AIC), McFadden’s R² were used to select the best-fit and best-specified logistic model. The variables of the logistic regression are described in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Description of variables.

https://doi.org/10.1371/journal.pone.0148586.t001

Results

Households with bed nets increased from 73.1% in the pre-intervention period to 87.1% post-intervention, and attained as high as 96.2% in intervention households, (Table 1). The estimated two-week occurrence of fever among under-fives declined from 56.6% at pre-intervention to 42.5% in the post-intervention period. Fever, however, was more prevalent in the control group than in the intervention group in both the pre- and post-intervention periods. The mean age of the mothers (caregivers) was comparable (about 27 to 28 years) across the study groups. Majority of the caregivers (78.9% and 74.6% in the pre- and post-intervention periods respectively) had no formal education. The mean age of the sampled children ranged from 2.2 to 2.6 among the study groups.

As shown in the contingency table (Table 2), the odd ratio (OR) of fever among the under-fives in households with bed nets relative to those without, was 0.783 [Confidence Interval (CI):0.67–0.95]. As such, the null hypothesis that the odds ratio is one was rejected; this implied that children who lived in households that possessed ITNs reported the incidence of fever 21.7% fewer times than children who lived in households without ITNs. The fever prevention fraction attributable to possession of bed nets in the household was statistically significant at OR = 0.217 (95% CI: 0.08–0.33). Similarly, children in the intervention group were significantly less likely to have malaria fever compared to children in the control group (OR = 0.765, 95% CI: 0.67–0.87) at the end of the study. The intervention prevention fraction was statistically significant at OR = 0.235 (95% CI: 0.13–0.33). The likelihood of fever in children post-intervention was also significantly less than the pre-intervention period (OR = 0.566, 95% CI: 0.50–0.65). The entire intervention strategy that was implemented significantly protected against malaria fever, (Prevention Fraction was OR = 0.434, 95% CI: 0.36–0.50).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Contingency Table of the Effect of Household Possession of Bed Nets, on the Risk of Fever in Children.

https://doi.org/10.1371/journal.pone.0148586.t002

The logistic regression in Table 3 showed that the estimated odds ratio of fever in children under-fives in households with ITNs and those in households without was 0.72 (95% CI: 0.60–0.88), an improvement from the estimated OR of 0.783 from the contingency analysis in Table 2. This indicates the odds of fever in under-fives in households with ITNs were significantly lower than the odds of fever in households without ITN and that bed nets offered 27.8% protection against malaria fever. The odds of fever in the under-five years old increased with the age of the mothers (or caregivers) (OR = 1.005, 95% CI: 0.997–1.01) although this was not significant. The odds ratio for households in the middle- and high- socio-economic levels relative to households in the low socioeconomic level, were (OR = 1.195, 95% CI: 1.03–1.39) and (OR = 1.179, 95% CI: 1.02–1.36) respectively. As such, the odds of fever in the under-five year olds that lived in middle and high socio-economic households were significantly higher than that of fever in low socio-economic households. The results also found that the educational levels of mothers (caregivers) influenced the odds of fever in under-five year olds. The odds of fever were lower with higher levels of mother’s education. The odds of fever in under-five year olds whose mothers had primary education did not differ significantly from those without formal education (OR = 1.007, 95% CI: 0.85–1.19). The odds of fever in the under-five year olds whose mothers had secondary education (OR = 0.69; 95% CI: 0.54–0.87) and tertiary education (OR = 0.45; 95% CI: 0.22–0.91) were significantly lower than for children of mothers with no formal education. There was a slight but significant decrease in the odds of fever as the number of the under five-year olds resident in households increased (OR = 0.995, 95% CI: 0.99–0.997).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Results of the Logistic Regression of Childhood Fever on Household Possession of Bed Nets, and Household Characteristics.

https://doi.org/10.1371/journal.pone.0148586.t003

The likelihood of fever in under five-year olds was significantly lower at post-intervention than at pre-intervention (OR = 0.39, 95% CI: 0.33–0.46) and this was the case with the treatment group compared to the control group (OR = 0.56, 95% CI: 0.48–0.66). The interaction between the treatment and period variable was also significant (OR = 1.703, 95% CI: 1.64–3.15).

Discussion

This study primarily estimated the impact of the interventions implemented to mitigate morbidity from malaria fever and explore how household characteristics influenced the impact of the interventions for reducing the prevalence of fever in the under five year olds in Bauchi State of Nigeria. These interventions consisted of two rounds of bed net distribution to households in all LGAs in the state on the one hand, and the integration and training of PMV into integrated management of childhood diseases including malaria, increased access to ACTs and training of service providers (through iCCM implementation) on the other. The results showed that household bed net coverage increased on average from 73.1% in the pre-intervention period to 87.1% post-intervention. Coverage reached 96.2% in households in the intervention groups at the end of the intervention period. The pre-intervention bed net coverage compared well with a previous coverage of 87% estimated in Bauchi State [15] in 2011. It was not clear why the intervention areas had better household bed net coverage, since all LGAs had a uniform average of 2.7 bed nets per household in 2015. A possible explanation suggesting that bed nets distribution in the first round could have been less in the control LGAs is farfetched. A study in Nigeria showed differences in net survival rates across regions [19]. Differing net survival rates between the study-groups may account for the differences in the coverage of bed nets found in this study. A more plausible explanation however, is that the observed differences in household coverage of bed nets was likely spin-offs from the activities of the trained Chemist, PMVs and SPs within health centers in the intervention areas as a consequence of the study.

The results showed that the two-week occurrence of fever among under-fives declined from 56.6% pre-intervention, to 42.5% post-intervention although the observed high prevalence of fever in the control group compared to the intervention group in both the pre- and post-intervention periods was statistically significant. Children that lived in households that possessed ITNs reported 21.7% less fever than children who lived in households without ITNs. Children in the intervention group were significantly less likely to have malaria fever relative to children in the control group, at the post intervention period. Being in the intervention group of the study significantly conferred on children, as much as 23.5% prevention against fever. The likelihood of fever at post-intervention was also significantly less than the pre-intervention period and protection fraction against fever attributable to the intervention, was statistically significant at 43.4%

The logistic regression showed the estimated adjusted odds ratio of fever in children under five years old in households with ITNs and those in households without was 0.722. Household possession of bed nets therefore provided about 27.8% protection against fever. The age of mothers (or caregivers) did not significantly predict the odds of fever in the under-five. The odds of fever in the under-five year olds living in the higher and the highest socio-economic households were 19.5%, 17.9% higher than the odds of fever in low socio-economic households. The results also showed that the educational levels of mothers (caregivers) influenced the odds of fever in the under-five year olds. Higher level of mother’s education significantly protected the children against fever. The odds of fever significantly decreased slightly with the number of the under five-year olds resident in households implying that children in households with high number of under-five year olds were slightly less likely to report fever than households with fewer children. Logistic regression also showed that the intervention significantly reduced occurrence of fever.

With regards to the influence of household demographic characteristics, on net ownership and use—which in this study, showed positive association between higher levels of mother’s education and high net ownership—a previous study in Bauchi State in 2011 similarly reported that mothers’ education and higher socioeconomic factors were positively associated with possession and use of bed nets [15]. Another study in Angola showed that children from wealthier households were less likely to be parasitaemic than those in poorest households. The poorest households in Tanzania, and Uganda were less likely to own a net. However, the poorest household in Angola had higher usage of bed nets than children in wealthier households [14]. In this present study, compared with mothers with no education, mothers’ education significantly protected against fever by a factor of as much as 31.2% in the children of mothers with secondary and by and 55.1% among children whose mothers had tertiary education respectively. Although the likelihood of fever was the same with the low and middle socioeconomic levels, the children in the highest socioeconomic group had 16.7% increased odds of fever relative to the low socioeconomic households. Households with higher socioeconomic profiles by implication have better economic access to factors of fever prevention and therefore, should experience lower prevalence rates of fever. Possible interactions between household possession of ITNs and the other unobserved factors may have contributed to this contrarian outcome. For instance, the Nigerian DHS of 2013 showed that possession of ITNs was more (61%) in rural households than in urban households (48%) [1]. Rural households however, were arguably less likely to be as socio-economically endowed as urban households were. So also, mothers with secondary and tertiary education are likely concentrated in the urban areas. The higher possession of ITNs in rural households despite lower household income relative urban households and a higher concentration of women with low to no education, largely reflect the subsidy policy of of Bauchi State government that delivered free ITNs targeting the poor and rural communities. It is likely that the subsidy program introduced market distortions in the possession of ITNs that obviated the expected positive relationship between socioeconomic status (household wealth index) and possession of ITN. In this respect, it can argued that the lack of association between household wealth and net ownership is an indication of the impact of the free bed net distribution in the mitigation of inequalities in household bed net ownership and in childhood malaria fever.

Overall, the interventions implemented in this study which included bed nets distribution and the integration of PMVs into integrated management of childhood malaria, significantly decreased the odds and significantly protected the children against fever. Increased bed nets coverage alone provided a significant 21.7% protection against fever whereas; the collective intervention provided as much as 58.1% protection. The impact recorded in this study demonstrate the potential in innovative public-private partnership for the reduction of common childhood diseases particularly malaria fever which is endemic in Bauchi state. The results underscore the evidence that operations of the Chemists and PMVs (including that of SPs) can be streamlined into providing integrated community case management of these childhood diseases particularly in malaria endemic communities where poor health systems and access to appropriate treatment are limited.

Conclusion

We conclude that interventions that concurrently combine bed nets distribution, with public-private partnerships with chemists and patent medicine vendors given appropriate training on treatment of fevers, training of SPs in health facilities, enhanced supplies and distribution of commodities, can be effective for reducing the prevalence and the likelihood of fever occurring in children under five years. Taken to scale, these combination of strategies can contribute significantly to the reduction of reduce the burden of malaria fever in the under-five year old children in Nigeria.

Acknowledgments

Data used in this analysis were generated from the United States Agency International Development (USAID) through the John Snow, Inc-led Targeted States High Impact (TSHIP) project in Nigeria. We are grateful to Bauchi State Ministry of Health and its agencies for the cooperation in giving access the fields work