Equity trends in ownership of insecticide-treated nets in 19 sub-Saharan African countries

Abstract Objective To examine the change in equity of insecticide-treated net (ITN) ownership among 19 malaria-endemic countries in sub-Saharan Africa before and after the launch of the Cover The Bed Net Gap initiative. Methods To assess change in equity in ownership of at least one ITN by households from different wealth quintiles, we used data from Demographic and Health Surveys and Malaria Indicator Surveys. We assigned surveys conducted before the launch (2003–2008) as baseline surveys and surveys conducted between 2009–2014 as endpoint surveys. We did country-level and pooled multicountry analyses. Pooled analyses based on malaria transmission risk, were done by dividing geographical zones into either low- and intermediate-risk or high-risk. To assess changes in equity, we calculated the Lorenz concentration curve and concentration index (C-index). Findings Out of the 19 countries we assessed, 13 countries showed improved equity between baseline and endpoint surveys and two countries showed no changes. Four countries displayed worsened equity, two favouring the poorer households and two favouring the richer. The multicountry pooled analysis showed an improvement in equity (baseline survey C-index: 0.11; 95% confidence interval, CI: 0.10 to 0.11; and endpoint survey C-index: 0.00; 95% CI: −0.01 to 0.00). Similar trends were seen in both low- and intermediate-risk and high-risk zones. Conclusion The mass ITN distribution campaigns to increase coverage, linked to the launch of the Cover The Bed Net Gap initiative, have led to improvement in coverage of ITN ownership across sub-Saharan Africa with significant reduction in inequity among wealth quintiles.


Introduction
Equity in health is a major tenet of global development organizations, such as the World Health Organization (WHO) and the World Bank, whose policies are aimed to decrease the gap between poor and rich populations. WHO defines health inequity as "inequality with respect to health determinants, access to the resources needed to improve and maintain health or health outcomes". 1 Many diseases, such as malaria, are not distributed equitably among populations. Malaria disproportionately affects poor, rural populations, with pregnant women and young children at highest risk of severe morbidity and mortality. [2][3][4][5][6][7][8][9][10] Addressing inequities that are actionable, such as the availability of commodities, has been the cornerstone of malaria control efforts for more than a decade.
In April 2008 the Roll Back Malaria Partnership, together with the Secretary-General of the United Nations, launched the initiative Cover The Bed Net Gap to achieve the goal of universal bed-net coverage by December 2010. 11, 12 The aim of this initiative is to have every household at risk of malaria transmission and every person within that household protected by an insecticide-treated net (ITN). 13,14 Since the launch of the initiative, countries have achieved high ITN coverage levels using various distribution channels such as community delivery, routine health services or outreach activities.
Before the launch of the initiative, many distribution strategies focused on populations at higher risk of malaria. The ITN policies frequently included distribution of nets to caregivers of children younger than five years during routine vaccination campaigns, distribution to pregnant women during antenatal care visits and via social marketing. In addition, ITNs could be purchased either at health facilities or in the private market. These distribution strategies led to inequity in ITN ownership among subgroups, 2,15 particularly between socioeconomic subgroups. Richer households were more likely to own ITNs than the poorest households, probably as a result of low access to health care among the poorest populations. 2,16,17 With the launch of the initiative, the distribution of ITNs shifted from targeted distribution to mass distribution campaigns. This shift gave malaria control programmes the opportunity to reduce disparities among subgroups by increasing ITN ownership [18][19][20][21][22][23][24] to reduce the malaria burden. 25 The mass distribution campaigns aim to provide one ITN for every two household members. Based on the longevity of the nets and the cost-effectiveness of conducting a mass distribution as compared to a targeted net replacement, these campaigns are recommended to take place every three years. 26 In 2015, seven years after the launch of the initiative, few multicountry studies have documented the effect of the mass distribution strategy on equity in ITN ownership coverage in sub-Saharan Africa. This study assesses the level of equity in bed-net ownership before and after the widespread implementation of national ITN distribution strategies.

Data
We used data from Demographic and Health Surveys (DHS) and Malaria Indicator Surveys (MIS), which are nationally representative, population-based household surveys and which use validated standard methods in all countries. Further details can be found elsewhere. 27 The analysis focused on malariaendemic countries in sub-Saharan Africa that have conducted DHS or MIS between 2003 and 2014. We defined surveys made before the launch of the initiative as baseline surveys Objective To examine the change in equity of insecticide-treated net (ITN) ownership among 19 malaria-endemic countries in sub-Saharan Africa before and after the launch of the Cover The Bed Net Gap initiative. Methods To assess change in equity in ownership of at least one ITN by households from different wealth quintiles, we used data from Demographic and Health Surveys and Malaria Indicator Surveys. We assigned surveys conducted before the launch (2003-2008) as baseline surveys and surveys conducted between 2009-2014 as endpoint surveys. We did country-level and pooled multicountry analyses. Pooled analyses based on malaria transmission risk, were done by dividing geographical zones into either low-and intermediate-risk or high-risk. To assess changes in equity, we calculated the Lorenz concentration curve and concentration index (C-index). Findings Out of the 19 countries we assessed, 13 countries showed improved equity between baseline and endpoint surveys and two countries showed no changes. Four countries displayed worsened equity, two favouring the poorer households and two favouring the richer. The multicountry pooled analysis showed an improvement in equity (baseline survey C-index: 0.11; 95% confidence interval, CI: 0.10 to 0.11; and endpoint survey C-index: 0.00; 95% CI: −0.01 to 0.00). Similar trends were seen in both low-and intermediate-risk and high-risk zones. Conclusion The mass ITN distribution campaigns to increase coverage, linked to the launch of the Cover The Bed Net Gap initiative, have led to improvement in coverage of ITN ownership across sub-Saharan Africa with significant reduction in inequity among wealth quintiles. and these were conducted between the years 2003-2008. Surveys carried out between years 2009-2014, we defined as endpoint surveys. All surveys are independent from each other.

Analysis
We did two sets of analyses: a country-level equity trend analysis of ITN ownership and a multicountry pooled analysis examining equity of ITN ownership by malaria transmission risk zones. Inclusion criteria for the country-level equity trend analyses included: (i) countries with one survey conducted between years 2003-2008 (baseline) and the other survey conducted between years 2009-2014 (endpoint); and (ii) all surveys must have included data on ITN ownership via a bed-net roster in the household questionnaire.
To explore if equity in ITN ownership varies by level of malaria transmission, the multicountry pooled equity analysis categorized all survey clusters into categories of malaria risk. We included surveys that had publicly available global positioning system (GPS) data for the surveyed clusters. If a country had more than one survey with GPS data in a time period, we used the most recent survey in both the baseline and endpoint. However, in Rwanda, we used the 2010 DHS GPS coordinates for the endpoint due to a lack of GPS coordinates for the most recent 2013 MIS survey. Only countries with GPS data from both surveys were included in the pooled analysis.

Defining ITN ownership
The outcome of interest is household ITN ownership, defined as the proportion of households with at least one ITN. As recommended by the Roll Back Malaria Monitoring and Evaluation Reference Group, the indicator is standard across countries and reflects the extent to which ITN distribution campaigns have reached all households. 28 For each survey, we calculated the proportion of households with at least one ITN. To test for significant changes in ITN ownership between baseline and endpoint surveys, we calculated 95% confidence intervals (CI).
We did not use the indicator for universal bed-net coverage, i.e. the proportion of households with at least one ITN for every two people, since the indicator was not launched until 2008 and therefore not captured in baseline surveys.

Defining wealth quintiles
The DHS wealth index measures economic well-being of households independently from health and education. 29, 30 The DHS wealth index is a survey-specific measure of the relative economic status of households based on analysis of household assets and service amenities at a particular point in time. Wealth quintiles (lowest, second, middle, fourth, and highest) ranking indicates relative rather than absolute economic status of the household. [30][31][32]

Defining malaria endemicity
We assigned each household cluster into geographical zones based on malaria transmission risk. To link DHS and MIS geo-coordinates (latitude, longitude) of each survey cluster to transmission risk zones, we used geo-coordinated Plasmodium falciparum parasite prevalence rates among children aged 2-10 years (PfPR 2-10 ) from the Malaria Atlas Project 2010. 33 We assigned all households in a cluster from the DHS or MIS survey  (13 851) in the no-risk category. We excluded clusters located in areas with no risk of malaria from analyses because populations in these areas would not be targeted by ITN distribution campaigns.
Due to small sample size in the low-risk group, we combined the intermediate and the low-risk groups.

Equity calculation
We used the Lorenz concentration curve (C-curve) and the Lorenz concentration index (C-index) to assess equity in household ITN ownership across household wealth quintiles. The Ccurve graphically presents the degree of economic-related inequality. 36,37 In the C-curve graphs, the x-axis presents the cumulative percentage of the sample, ranked by wealth, beginning with the poorest, while the y-axis presents the cumulative percentage of the variable of interest corresponding to the cumulative percentage of the distribution of wealth. 36 In the C-curve graphs, the dashed 45° line represents equity whereby the health outcome is distributed equally among all wealth quintiles. The C-curve will be below the equity line if ITN ownership is concentrated in richer households and will be above the equity line if ITN ownership is predominantly among poorer households.
The C-index, which provides quantification of this measure of equity, is defined as twice the area between the C-curve and the 45° line of equity.
We calculated the C-index values by using the following equation where P is the cumulative percentage of the household ranked by economic status in group t, and L is the corresponding concentration curve ordinate. 37

Research
Ownership of insecticide-treated nets in sub-Saharan Africa Cameron Taylor et al. 0 suggests no difference in ITN ownership between different wealth quintiles. A C-index larger than 0 suggests that ITN ownership is predominantly among the richer households. Conversely, a negative index indicates that ITN ownership is more concentrated among the poorer households. 17,38 We used the concindc command in Stata version 13 (StataCorp. LP, College Station, United States of America) to calculate the C-index values and their standard errors and we used the clorenz command for producing the C-curves. We calculated 95% CI for the C-index values.

Results
In total, 19 countries (45 surveys) met the inclusion criteria for country-level equity analysis. For the multicountry pooled equity analysis, we included 15 countries (30 surveys; Table 1).

ITN ownership
In all countries, except Angola, there was a statistically significant increase in ITN ownership between the baseline and endpoint surveys. Rwanda and the United Republic of Tanzania showed the greatest improvement in ITN ownership, from 15% to 83% and from 23% to 91%, respectively. Angola displayed the smallest improvements in ITN ownership (from 28% to 35%; Fig. 1).

Country-level equity analysis
At the country level, 13 out of the 19 countries showed improvements in equity of ITN ownership between baseline and endpoint surveys, while two countries showed no changes and four countries displayed worsened equity ( Table 2).
For all countries showing improvements in equity, the ITN ownership was concentrated in households from the highest wealth quintiles in the baseline surveys, as indicated by a C-curve below the equity line (C-index > 0). However, the countries showed different levels of improvement. For Burkina Faso, the Democratic Republic of the Congo, Malawi, Rwanda and Uganda equity significantly improved with C-index values closer to zero in the endpoint surveys. For Benin, Cameroon and the United Republic of Tanzania, equity in ITN ownership across wealth quintiles (C-index = 0) had been achieved. The endpoint surveys from the Congo, Guinea, Nigeria, Sierra Leone and Zimbabwe showed higher levels of ITN ownership among poorer households (C-curve above the equity line and C-index < 0; Fig. 2; Table 2).
In Mali, ITN ownership was equally distributed across wealth quintiles in both baseline and endpoint surveys with no significant change. In Mozambique, ITN ownership remained concentrated among richer households in the endpoint survey. However, C-index values were close to zero ( Fig. 2; Table 2).  12 Years of the surveys are given in Table 1. A value of 0 suggests no difference in ITN ownership between different wealth quintiles. A C-index larger than 0 suggests that ITN ownership is concentrated among the richer households. Conversely, a negative index indicates that ITN ownership is more concentrated among the poorer households. 17,38 Ownership of insecticide-treated nets in sub-Saharan Africa Cameron Taylor et al.
Madagascar and Senegal maintained levels of inequity that favoured the poorest households. In Angola and Niger, while the inequity in the baseline surveys was close to zero, in the endpoint surveys, household ITN ownership increased and was in favour of the richer households ( Fig. 2; Table 2). Fig. 3 shows a scatter plot of the Cindex by ITN ownership for all surveys included in the country-level analyses. The plot indicates a decline in the disparity of the C-index values as ITN cov-erage increases. Surveys that took place between 2009-2014 have higher levels of ITN ownership and greater equity compared to surveys from 2003-2008.

By transmission risk
In high malaria transmission risk zones, ITN ownership was concentrated in households from the higher wealth quintiles in the baseline surveys (C-index: 0.07; 95% CI: 0.06 to 0.08). However, in the endpoint surveys this inequity was no longer evident (C-index: 0.00; 95% CI: 0.00 to 0.01; Fig. 4). In the

Research
Ownership of insecticide-treated nets in sub-Saharan Africa Cameron Taylor et al.
low and intermediate malaria transmission risk zones, the ITN ownership was in favour of richer households in the baseline surveys (C-index: 0.14; 95% CI: 0.13 to 0.14), but shifted to favour the poorer households in the endpoint surveys (C-index: −0.01, 95% CI: −0.02 to −0.01; Fig. 4).

Discussion
This study presents evidence of the positive impact of mass ITN distribution strategies on equity of ITN ownership in 19 malaria-endemic countries in sub-Saharan Africa. In 15 of the 19 countries analysed, ITN ownership either became more equitable or maintained equity between baseline and endpoint surveys. In four countries, improvements in equity could not be detected between the two surveys: in two countries the ITN ownership remained concentrated among households from the lowest wealth quintiles, while in the other two, ITN ownership remained concentrated among the richer households.
The pooled multicountry analyses further supported the findings that the significant increase in ITN ownership has favoured the poorest households in most settings. In the countries where ITN ownership either became more equitable or maintained equity, all showed a significant increase in the proportion of households with at least one ITN between baseline and endpoint surveys. In countries with very high levels of household ITN ownership, such as Rwanda and the United Republic of Tanzania, the chances of equitable distribution are inherently higher. Thus, the recent funding for malaria control 25 and the subsequent investment in mass ITN distribution campaigns have likely contributed to reduced inequity among wealth quintiles by increasing overall coverage.
Angola, Mali, Mozambique and Niger had close to equitable ITN ownership in their baseline surveys despite relatively low overall levels of ITN ownership. This finding could be due to early implementation of focused nationwide campaigns where ITN distributions were integrated with child health campaigns. 19,[40][41][42] In Mali and Mozambique, ITN ownership remained equitably distributed in endpoint surveys, possibly due to the rollout of net-distribution campaigns to achieve universal coverage in 2008 and 2011. 43,44 However, Angola and Niger experienced a decrease in equity in their endpoint surveys despite low inequity at baseline. The trend of increasing inequity in Angola could be partially due to the timing of campaigns in relationship to the survey as well as a shift in ITN distribution from integrated campaigns to distribution in only selected municipalities. 45 Reasons for the worsened equity in Niger is less clear, but possible explanations could include the lack of implementation of free ITN distribution campaigns between baseline and endpoint surveys. 19,25 Madagascar and Senegal were the only countries that maintained levels of inequity from base-line to endpoint in favour of households from the lowest wealth quintiles. The trend seen in Congo, Guinea, Nigeria, Sierra Leone and Zimbabwei.e. ITN ownership shifted from being concentrated in the richer households to being concentrated in the poorer households -is not surprising. These countries have moved towards universal ITN coverage for populations at-risk and shifted their distribution of ITNs to high-risk rural-areas, which are usually less wealthy than urban centres. 2 Another reason for this trend may be that wealthier households have access to a wider range of other effective interventions, such as improved housing with screened windows and doors and closed eaves that make ITNs less essential for malaria prevention.
To explore if equity of ITN ownership varies by malaria transmission risk, we pooled clusters into two groups stratified by low/intermediate and high levels of malaria transmission. In the pooled analysis, equity increased significantly in both groups. However, the greatest improvement in equity occurred in clusters in low-and intermediate-risk zones. The observed results could be due to changing ITN policies between baseline and endpoint surveys, and more specifically, the rollout of free mass distribution campaigns after 2008. Before 2008, financial and logistic constraints caused most distribution campaigns to be targeted to high-risk populations (children younger than five years and pregnant women) and/or high-risk regions (rural, high-transmission zones). Therefore, households from the lowest wealth quintiles in low-and intermediate-risk zones were less likely to own a net if they did not have access to healthcare services or could not afford to pay for a net at market price. The shift to free mass distribution campaigns may have improved equity by providing access to the households from the lowest wealth quintiles that did not have previous access to nets.
This study has a few limitations that should be highlighted. The wealth quintiles are based on assets, which may be different from country to country as the individual assets might have different weights in the principal component analysis. While the analysis used data from 19 countries, other countries may have experienced changes in equity, but were not captured in this analysis. In addition, we excluded four of the countries

Research
Ownership of insecticide-treated nets in sub-Saharan Africa Cameron Taylor et al. used in the country-level analysis from the pooled analysis due to a lack of GPS data. This study focused on equity of ITN ownership and did not assess ITN use or ITN access, i.e. the proportion of the population who could use an ITN with the assumption that one ITN can protect two individuals. Future studies should examine equity of ITN access as it is a more comprehensive measure of the level of protection within a household. With more countries implementing universal bed-net coverage strategies, capturing changes in equity of ITN access through survey data will be possible.

Research
Ownership of insecticide-treated nets in sub-Saharan Africa Cameron Taylor et al.