Estimating the Number of Insecticide-Treated Nets Required by African Households to Reach Continent-wide Malaria Coverage Targets FREE

Use of insecticide-treated mosquito nets (ITNs) for prevention of malaria in Africa is one of the pillars of the Roll Back Malaria (RBM) partnership, especially for children younger than 5 years and pregnant women, populations that are particularly vulnerable to malaria. In trials, ITNs have been shown to reduce all-cause mortality in children younger than 5 years across malaria-endemic countries in Africa.^1- 2 In addition, ITNs have been shown to reduce morbidity and adverse birth outcomes in pregnant women.^3- 5

Monitoring ITN coverage in African countries is central to evaluating the progress of malaria control under the RBM partnership. At the African Summit on Roll Back Malaria in April 2000, African heads of state committed themselves to ensuring that by 2005, at least 60% of those at risk of malaria, particularly pregnant women and children younger than 5 years, would benefit from the protection of ITNs.⁶ By 2005, very few African countries reported progress toward these targets,⁷ despite which the World Health Organization agreed to raise coverage to 80% by 2010.^8- 9 Although progress has been slow, achieving these targets is aided by increased funding and operational support provided through RBM partners. In particular, initiatives such as the Global Fund to Fight AIDS, Tuberculosis and Malaria (GFATM), the US President's Malaria Initiative, the World Bank Malaria Booster Programme, integrated-ITN campaigns through the Measles Initiative, and the Bill and Melinda Gates Foundation–supported Malaria Control and Evaluation Partnership in Africa (MACEPA), along with national governments, are accelerating ITN distribution to achieve high coverage in several areas.^10- 15 Progress is also reinforced by improvements in net manufacturing technology, thanks to which conventional ITNs that required at least annual re-treatment are being replaced with more durable and longer-lasting insecticidal nets (LLINs), whose residual insecticidal activity lasts throughout their expected life span^16- 18 without need for re-treatment.

Two indicators are used commonly for monitoring mosquito net coverage: (1) household possession of nets, which indicates the extent to which distribution channels are enabling high coverage, and (2) use of nets by children younger than 5 years, which correlates with protection and, therefore, is a more direct predictor of epidemiological impact.¹⁹ Between 1999 and 2004, nationally representative household surveys provided the first coverage assessments for 36 sub-Saharan countries in Africa. From these surveys, only 3% of children younger than 5 years and 2.8% of pregnant women slept under ITNs.⁷ These unacceptably low coverage estimates are partially explained by very low availability of nets within African households, with only 18% of households possessing 1 or more nets (insecticide-treated or untreated).

The purpose of this article is to estimate the total number of nets and ITNs currently available at the household level in sub-Saharan Africa and the number needed to achieve the 80% coverage target for children younger than 5 years and pregnant women. We compared survey data on current net and ITN coverage with estimates of population and households at malaria risk.

The primary sources of data on mosquito net coverage used in the analysis were the Multiple Indicator Cluster Surveys II (MICS)²⁰ and the Demographic and Health Surveys (DHS),²¹ 2 well-established international household survey instruments used in many developing countries for monitoring and evaluating a variety of health interventions. Between 1999 and 2001, MICS surveys conducted in 24 malaria-endemic African countries, with support from the United Nations Children's Fund, included questions on the use of mosquito nets and ITNs among children younger than 5 years. Multiple Indicator Cluster Surveys are nationally representative household surveys, sampled through a 2-stage cluster design with a sample size of approximately 5000 households. The DHS surveys are nationally representative household surveys that focus on reproductive and child health. Demographic and Health Surveys are organized through national ministries of health, national statistics bureaus, and the MEASURE DHS program as implemented by ORC Macro, Calverton, Md. Typically, DHS consist of interviews with approximately 8000 women aged 15 to 49 years living in households that are sampled in a multiple-stage cluster design.

The survey instruments have been extensively validated through pretesting between 1998 and 2003 and further standardized through international collaboration coordinated through the RBM Monitoring and Evaluation Reference Group (MERG) Household Survey Task Force. These instruments represent the gold standard for assessing global targets for ITN coverage and are used by most countries for reporting on the United Nations' Millennium Development Goals.²² Between 1998 and 2006, 22 DHS surveys in African countries asked questions related to household net possession or use.

Where surveys were not available from DHS or MICS, we considered other nationally representative estimates or malaria-specific surveys assessing net and ITN coverage, which included recent surveys from Eritrea, Somalia, Togo, and Zambia.^23- 26 The May-June 2006 survey in Zambia was the first application of the standardized RBM MERG Malaria Indicator Survey, following the DHS questionnaire model, which assesses coverage of malaria interventions along with malaria parasite and anemia prevalence.⁷ Final reports from the MICS, DHS, and RBM MERG Malaria Indicator Surveys include details of response rates and estimate precision, and the results reflect weighted, nationally representative estimates to account for design effect of cluster and household selection.

The total number of mosquito nets (treated or untreated) was calculated from the proportion of households that possessed at least 1 net or ITN, the average household size, the number of households in the country, and the average number of nets or ITNs per household. The proportion of households that possessed at least 1 mosquito net (irrespective of whether insecticide-treated) was known from 21 nationally representative or other surveys; similar data for ITNs specifically were available from 17 surveys. For the remaining countries with survey data available, household possession was estimated from reported use by children younger than 5 years, based on a previously established correlation for African countries between 1999 and 2002¹⁹: (1) percentage of net use in children younger than 5 years = −4.2% + 0.875 × (percentage of households possessing 1 or more nets) (R² = 0.89; P<.001) and (2) percentage of ITN use in children younger than 5 years = 0.55 × (percentage of households possessing 1 or more ITNs) (R² = 0.97; P<.001).

For countries with no estimate of possession or use in children younger than 5 years, we used as a nearest-neighbor approximation the average from possession estimates (either directly observed or estimated from net use) from all neighboring countries with a survey estimate available. For countries where surveys had measured net possession but not ITN possession, the latter was estimated from the former using the mean ratio in possession of nets vs ITNs from all surveys that had measured both of these indicators.

The total number of households within a country was estimated from average household size and total population. For 34 countries, the most recently available nationally representative survey provided the average household size. For countries with no available estimate of average household size, the median value of documented median household sizes from these 34 other countries (5.1) was used. Total population, population younger than 5 years, and female population aged 15 to 49 years were abstracted from the World Population Prospects Population Database for each country in Africa, for the year in which the corresponding net coverage survey was conducted or the year 2003.²⁷ The medium variant assumptions of changing mortality, fertility, and migration for total population size were used.

Not all households in sub-Saharan Africa are at risk of malaria, nor do all households have pregnant women or children younger than 5 years. The proportion of households with children younger than 5 years and with women aged 15 to 49 years was reported in 22 MICS surveys. We estimated the proportion of households with pregnant women from the proportion of households with women aged 15 to 49 years and the ratio of pregnant women to all women aged 15 to 49 years. The latter number was derived from numbers of pregnancies, estimated from numbers of live births (projection for interval 2000-2005 from the World Population Prospects Population Database²⁷ plus the number of maternal deaths during pregnancy^27- 28). For countries with no MICS data available, the median value of all other countries (ie, 61.8% of households with children <5 years and 86.7% with women aged 15-49 years) was used.

The aforementioned relationship between household ITN possession and use among children younger than 5 years¹⁹ was further used to derive the total required number of ITNs from the resulting total households.

Estimates of populations living in areas at risk of malaria transmission were derived from the Mapping Malaria Risk in Africa (MARA/ARMA) project.²⁹ The definition of a population at risk of malaria was based on a continental distribution model of climate suitability for malaria transmission that, for all countries in middle Africa, produced the following risk categories: no malaria risk, epidemic malaria risk, and endemic malaria risk.³⁰ Southern African countries (Botswana, Namibia, South Africa, Swaziland, and Zimbabwe), which have a different pattern of malaria transmission, were categorized as areas of any malaria risk or no malaria risk.³¹ MARA combined this model with a population distribution map from interpolated human settlement to determine the number of people at any malaria risk (which includes both epidemic and endemic malaria) or endemic malaria risk.³² The population totals were aggregated to the national level to determine a fixed percentage of population at risk in each country.³³

To apply the national-level coverage estimates to percentage populations at risk, we assumed that currently available nets are equally distributed across risk and nonrisk areas. This was thought to be reasonable because in many surveyed countries, net coverage is equally high or higher in urban (less malarious) areas than in rural (more malarious) areas.³⁴ The sensitivity of outcomes to this assumption was assessed in a sensitivity analysis.

Table 1 presents the reported and estimated household possession of any nets and ITNs by country. Of the 48 countries in sub-Saharan Africa, 43 were included in the analysis. Djibouti, Cape Verde, Lesotho, and the island nations of Seychelles and Mauritius were excluded because the risk of malaria transmission is minimal, absent, or confined to small focal areas. Sudan was considered part of sub-Saharan Africa. Twenty-three countries had nationally representative or multidistrict survey estimates (18 DHS, 1 AIDS Indicator Survey,³⁵ and 4 malaria-specific^23- 26) of household possession of nets between 1999 and 2006 that were included in the analysis; 19 countries had nationally representative or multidistrict survey estimates (15 DHS and 4 malaria-specific surveys) of household possession of ITNs. A further 16 countries had nationally representative surveys (15 MICS and 1 DHS) measuring use among children younger than 5 years of nets and ITNs, from which we estimated household possession. For countries where multiple surveys were available, preference was given to estimates based on a nationally representative sample design and to those that were more recent. For 3 countries (Botswana, Liberia, and South Africa), no estimate of either net or ITN possession or usage was available, and we used a nearest-neighbor approximation.

Across the 43 countries in sub-Saharan Africa, the estimated overall population-weighted proportions of households that possessed any net(s) or ITN(s) were 23.8% and 6.7%, respectively (Table 1). For the median survey year 2003, a total of 31.9 million households were estimated to have at least 1 mosquito net and 9.4 million households to have at least 1 ITN in sub-Saharan Africa.

To convert households possessing 1 or more nets (or ITNs) to total number of nets, we multiplied the reported households with 1 or more nets by the average number of nets per household. Of the available nationally representative surveys, 15 included questions on the average number of nets or ITNs a household possessed. From these surveys, a clear correlation emerged between the average number of nets and ITNs per household and the proportion of households with at least 1 net or ITN (Figure 1). This relationship was subsequently used to estimate the average number of nets and ITNs in households for surveys that had not directly measured the ITNs. For countries with no measure of percentage of households with at least 1 ITN, this measure was estimated using the average ratio in possession of nets vs ITNs observed from surveys that had measured both of these indicators. Multiplying the total number of households and the estimated number of nets and ITNs per household for each country, a total of 57.8 million nets and 18.4 million ITNs were estimated to be present in households in these countries (Table 1). Of these, we estimated that 16.7 million ITNs were located in areas at any malaria risk and 15.2 million ITNs were located in areas at endemic risk (Table 2).

For the median survey year 2003, we estimated that in sub-Saharan Africa, 133.4 million children younger than 5 years and pregnant women are living at risk of malaria; 118.5 million of these live in areas of endemic risk (Table 2). These children younger than 5 years and pregnant women live in an estimated 86.4 million households at risk of malaria, of which 75.0 million households are at endemic risk.

The required number of ITNs for use by 80% of target populations does not directly follow from these population sizes, since this depends on behavioral usage patterns: are all owned nets in use and, if so, by pregnant women and/or children younger than 5 years rather than by other household members? How many women and young children sleep under the same net? Also, the extent to which net distribution can be targeted to households in risk areas (how many owned and used nets are in households at actual malaria risk) and, more specifically, to households with pregnant women and children, will make a difference. A previous analysis of 12 African DHS surveys found that, on average, in only 55% of households that possessed 1 or more ITNs were children younger than 5 years (among) the reported net users (R² = 0.97; P<.001).¹⁹

If we assume that all ITNs could be effectively targeted to households with pregnant women and children younger than 5 years in areas at risk and that 55% of nets owned by these households would be used by those target groups, at a minimum 119.4 million ITNs would be needed to ensure that 80% of pregnant women and children younger than five years at risk slept under them during the median survey year 2003. In the absence of effective targeting to households with young children and/or pregnant women, this same usage target would require 175.2 million ITNs for 2003. Assuming instead that ITNs will get equally distributed over areas at risk and not at risk, the number required for use by 80% of young children and pregnant women would be 203.9 million. And, as a maximum, a fixed 2 ITNs for all households in risk areas, regardless of whether they have pregnant women and/or children younger than 5 years, would require a total of 240.9 million ITNs.

The minimum, mean, and maximum estimates of 119.4 million, 175.2 million, and 240.9 million required ITNs from the median survey year 2003 used in the analysis would correspond to 130.2 million, 192.1 million, and 264.1 million ITNs in 2007, respectively, given population growth (Figure 2). Assuming that nets wear out after 4 years,^18,36 these numbers would correspond to at least 37 million additional ITNs every year to achieve the 80% ITN coverage target for the African continent by 2010. Figure 2 also shows what increase in net availability might be expected over time, based on the observed increase between 2000 and 2003.

We assessed the sensitivity of estimated numbers of nets and ITNs present in households against several assumptions (Table 3).

Under the assumption that net distribution has until now been effectively targeted so that all currently available mosquito nets and ITNs are present in malaria risk areas, the estimated numbers of nets and ITNs in areas at any risk of malaria increase slightly, from 53.6 million to 57.8 million and from 16.7 million to 18.4 million, respectively.

Not all surveys identify a treated net in the same manner. Recent DHS surveys (≥2005) defined ITNs as those obtained or re-treated with insecticide in the 12 months preceding the survey, whereas most DHS surveys conducted in 2003 or 2004 reported nets as insecticide-treated only if they were obtained or re-treated in the 6 months preceding the survey, a more strict definition. The revised, less strict definition corresponds better with operational re-treatment efforts, such as those conducted during focused annual child health weeks or with mass immunization campaigns, as are common in many of these countries. Surveys with the new, less strict definition had, on average, about 0.16 more ITNs per household. While this may be a result of country-specific distribution success or timing of measurement, adjusting results from surveys reporting the more strict definition by this difference leads to a higher estimate: 22.1 million ITNs available within all malaria risk areas. Using a fixed median number of nets per household (0.44 for any net and 0.13 for ITNs; Table 1) instead of country-specific estimates did not greatly affect the total number of nets and ITNs in malaria risk areas (estimates 53.6 and 16.2 million, respectively).

Availability of ITNs in South Africa, Botswana, and Liberia, countries with no available survey reports of household possession of nets or ITNs, may in reality be lower that what we assumed based on data from neighboring countries because of recent internal conflict and health service delivery interruption in Liberia³⁷ and the relatively low burden of malaria and reliance on indoor residual spraying in South Africa and Botswana.³⁸ Changing these assumptions did not, however, appreciably affect the continental estimates (Table 3).

Achieving high ITN coverage, defined as 80% of children younger than 5 years and pregnant women sleeping under an ITN the night before the survey interview, requires many more ITNs than are currently available in African households. Based on available survey data between 1999 and 2006, we estimated that 18.4 million ITNs are available in households in 43 countries in sub-Saharan Africa; of these, 16.7 and 15.3 million are likely to be present in areas at any malaria risk and endemic malaria risk, respectively. Using evidence that links ITN availability at the household level to ITN usage among target populations provides a more realistic estimate of the numbers of ITNs required to reach high continent-wide coverage targets. Furthermore, it guides operational ITN delivery requirements to ensure a maximum health impact.

Several limitations in coverage data make this first continent-wide estimate uncertain. Of particular concern, the surveys in Somalia and Sudan did not include representative national samples. In Somalia, restricted access due to security concerns makes comprehensive household surveys difficult, while in Sudan, only the northern, less malarious area of the country was included. Estimates will be improved and updated with forthcoming DHS and MICS surveys in additional countries that use a more standardized malaria survey questionnaire, method of analysis, and tabulations.^20- 21 In addition, the RBM MERG promotes a standardized household survey method for consistent, cross-country comparison as the scale-up of malaria interventions progresses.⁷ This method, based on a full net roster within households, identifies the range of ITNs available including conventional nets that were recently obtained and/or re-treated within the previous 12 months, as well as LLINs, all of which are considered insecticide-treated.

Further uncertainty relates to biases in the interpretation of survey data. Many DHS and MICS surveys were conducted in the dry season, when mosquito net use is typically lower (up to 5-fold^39- 42) than during the rainy season, when most malaria transmission occurs. For countries where we estimated net possession from net use, we may thus have underestimated the number of available nets.

For ITNs specifically, many older surveys did not inquire whether ITNs had been obtained or (re-)impregnated within the preceding 6 or 12 months, which is, however, critical to maintain efficacy in killing and repelling mosquitoes. At the time of many of the surveys analyzed here, when large-scale ITN distribution was just beginning, most ITNs were likely to have been recently acquired and impregnated. However, for future DHS, MICS, and malaria-specific surveys, it will be important to include questions on time since reimpregnation (and washing, which influences insecticide disappearance⁴³) and branding, to identify LLINs.

In addition to nets already present within households, for which household surveys measure coverage, countries have increasing numbers of nets available in the distribution systems or in the marketplace, which had not yet reached households. Notably, since 2003 almost all African countries have received or will receive money for malaria control from the GFATM through its first 5 rounds of grant proposals.¹⁰ These GFATM grants, the largest committed funding source enabling ITN procurement, include allocations for the procurement of approximately 109 million ITNs worldwide, of which by the end of December 2006 approximately 18 million had been delivered to households.¹⁰ The remaining ITNs are awaiting purchase through tendering and procurement processes, are in the production pipeline, or are awaiting distribution at national or local levels over the 5-year terms of the respective grants. In addition, the World Bank Malaria Booster program supports 9 African countries, with an anticipated expansion to 6 more, with support for ITN procurement embedded in many countries' action plans.¹⁵ Activities of the US-based President's Malaria Initiative, including support for ITN procurement, are currently targeted in 7 African countries.¹¹ This concentration of funding provides realistic opportunities for countries to deploy ITNs on a large scale and to sustain high coverage over time.

The introduction of new ITN products will also transform the availability and measurement of ITN coverage. To date, only 3 ITN products have been granted LLIN status, thus conforming to the World Health Organization Pesticide Evaluation Scheme 20-wash criteria, and are identified through their branding in surveys.⁴⁴ A new product to transform conventional nets into LLINs, using K-O Tab 1-2-3 tablets (Bayer Vector Control) has shown promise in evaluations and is gaining momentum to address the needs of existing net re-treatment.^44- 45 For understanding insecticidal status of nets, this presents new challenges for measurement 12 months after the last reported re-treatment.

Coverage with ITNs is clearly increasing, given that ITNs did not exist outside of research settings until 1997. The pace of increase in net coverage observed between 2000 and 2003 (Figure 2) suggests that coverage in 2005 was unlikely to have achieved the then 60% African Summit on Roll Back Malaria target and that achievement of 80% coverage by 2010 would require a substantial acceleration of net distribution efforts. However, more rapid increases in net coverage have been documented in specific ITN promotion programs recently, particularly in Togo, Burundi, Tanzania, and Ghana,^{26,41,46- 47} indicating that with increased financial and human resources it is possible to rapidly scale up ITN coverage. The impressive increases in (budgeted) ITN procurements achieved during the past 3 years through international financing initiatives further justify the impression that ITN coverage is increasing exponentially.

Regular representative household surveys of ITN possession and use are needed in all countries to enable comprehensive assessments and evaluation of progress over time and in a timely manner. The 2006 round of MICS and several ongoing DHS surveys are due for release by mid-2007, providing much needed progress benchmarks for many African countries.

The need for ITNs is dramatic. Between 2003 and 2010, at least 37 million additional ITNs will be needed in households annually to achieve the 80% ITN coverage target for the African continent by 2010, assuming available nets wear out after 4 years^18,36 (Figure 2). With modest cost estimates for long-lasting ITN purchase (US $4.55)⁴⁸ and delivery (US $1.40),⁴⁹ this represents more than US $220 million annually to increase household availability of ITNs from existing levels through rapid scale-up channels. With available funding, many donor “focus” countries will likely achieve high coverage by 2010, but gaps will remain in many countries, especially in regard to maintaining coverage after initial rapid scale-up targets are met.

In addition to more ITNs, efforts are needed to promote good net usage patterns and regular insecticide (re-)treatment among target populations, especially in areas where levels of household net possession are already high. In 2003, the vast majority of ITNs available in households were not LLINs, but conventional nets that require re-treatment at regular intervals to maintain ITN status. Over the short term, net re-treatment efforts are needed to sustain ITN coverage levels and help bridge the gap between available treated and untreated nets. As more nets are replaced with long-lasting ITNs, the proportion of available nets that are ITNs will increase and reduce the need for net re-treatment. The volume of ITNs and re-treatments will require coordination among donor agencies, public health officials, net manufacturers, and existing distributions systems to ensure timely delivery and retreament to achieve the agreed targets.