Rolling out insecticide treated nets in Eritrea: examining the determinants of possession and use in malarious zones during the rainy season

Introduction

The use of insecticide treated bed nets (ITNs) to protect vulnerable populations from malaria parasite transmission is one of the main strategies recommended by the Roll Back Malaria (RBM) partnership (WHO 2002). While the protective efficacy, effectiveness and cost effectiveness of ITNs have been well established (Choi et al. 1995; Goodman et al. 1999; Schellenberg et al. 2001; ter Kuile et al. 2003a,b,c; Meltzer et al. 2003; Phillips-Howard et al. 2003; Wiseman et al. 2003; Lengeler 2004), many challenges remain as national programmes begin to scale up coverage. Such challenges include: access and availability of ITNs, cost of ITNs on the open market in resource poor settings, timely re-impregnation of bed nets and issues of proper ITN adherence, deployment and use.

Recently, a number of studies have described local perceptions of the acceptability of bed nets and insecticide, as well as the determinants of ITN possession and use (Winch et al. 1994, 1997; Binka & Adongo 1997; Agyepong & Manderson 1999; Schellenberg et al. 2001; Alaii et al. 2003). Socio-economic factors such as wealth, access to health care and education have been shown to be important predictors of ITN possession and use (Winch et al. 1997; Schellenberg et al. 2001; Heggenhougen et al. 2003). Ethnicity has also been reported as an important factor, as people with pastoralist and semi-nomadic lifestyles may be less likely to possess and use an ITN as compared to settled agricultural communities (Bradley et al. 1986; MacCormack & Snow 1986; Aikins et al. 1993; Thomson et al. 1996). Some research also suggests that gender may influence the use of ITNs within households, as different roles dictate different sleeping patterns for men and women (Aikins et al. 1993). Additionally, several investigations in urban and peri-urban Africa suggest desire for mosquito avoidance is a strong determinant of net usage (Binka & Adongo 1997; Agyepong & Manderson 1999; Macintyre et al. 2002). This is, of course, strongly influenced by season, but is also an effect of correct understanding of the risk of transmission.

Surprisingly, few published analyses have differentiated between possession and use, and yet this distinction is fundamental. For even though households may report they possess an ITN, if the net is not at least hung up during peak transmission seasons, its efficacy may be zero. Thus, while coverage of household ITN possession may be an important indicator for managers, it says little about the likely epidemiological impact of the programme. A recent assessment of several national surveys suggests there is considerable disparity between ITN possession and use (Korenromp et al. 2003). ITN possession was shown to range between 0.1% and 28.5%, while use among children less than 5 years old ranged between 0% and 16%. This disparity between household ITN possession and individual use may include such factors as: the rationing of ITNs as a result of there being fewer nets than members or beds in a household, issues related to local understanding of transmission and mosquito avoidance behaviours.

While challenges to ITN ownership may be eventually overcome as a result of the expansion of large-scale distribution efforts, and as social marketing campaigns increase the perceived benefit of ITNs and subsequently increase demand, the question of who is and who is not using ITNs once they are made available remains unanswered. Outstanding issues related to ITN use at the household and individual levels include determinants of proper net deployment as well as individual adherence. The issue of how adherence varies as a function of season is also extremely important as the use of ITNs only when mosquitoes are perceived as a nuisance, or only when the weather is cool (i.e. during the rains) (Alaii et al. 2003), may place many individuals at risk of malaria infection outside the immediate rainy season.

Answers to such questions are important and must be fed back to programme planners and policy makers to achieve the greatest impact of ITN programmes. This paper examines some of the social, economic and locational factors effecting household ITN possession and use in the context of Eritrea's National Malaria Control Programme (NMCP). We explore ‘use’ of an ITN in two ways: first, we define use as a proportion of all households in the survey with either all members under an ITN or all children less than 5 years old under an ITN the night before the survey. This is the definition recommended by RBM for measuring the proportion of households in a survey's universe whose members are protected by an ITN. Second, we examine the determinants of ITN use among only those households who already own at least one ITN. This second outcome variable measures more intuitively the gap between ownership and post-ownership use. The intention of this analysis is to examine the influences on both possession and use and to extract lessons applicable to the Eritrean context, or more broadly to other programmes that are using ITNs to reduce the burden of malaria.

Methods

Study area

Eritrea, with an estimated population of 3.5 million people, is divided into six administrative zobas (zones). Altitudes range from below sea level to 3000 m. Temperatures vary widely, although the western lowlands and coastal plains are generally associated with extremely hot and dry climatic conditions. Precipitation ranges from 200 mm/year in the lowland areas to 1000 mm/year in the highlands (Macro 2003). Anopheles arabiensis mosquitoes, belonging to the A. gambiae complex, are the main vectors in the country (Shililu et al. 2003a,b, 2004). Plasmodium falciparum is the primary species of malaria found in Eritrea, although P. vivax has also been reported (Shililu 2001). Malaria transmission is seasonal, with estimates of entomological inoculation rates ranging from 0 to 70.6 infective bites per person per year (Shililu et al. 2003b). Malaria accounts for approximately 28% of all hospital admissions and 30% of all outpatient morbidity, although malaria-free areas in the highlands do exist (Shililu 2001).

The data presented here were collected in September and October of 2003 in three zobas of Eritrea: Anseba, Gash-Barka and Debub (Figure 1). According to the NMCP, these zobas are the major malarious areas of the country, and thus also correspond to areas where intense malaria control efforts have been initiated. Since 2002, control efforts have included the distribution of free pre-treated ITNs for all households in malarious areas (a minimum of two nets per household was the goal for the programme managers in 2002–2003), improved malaria case management, indoor residual spraying (IRS) in selected villages and larval control and environmental management (again, in selected areas). The nets are distributed by ministry personnel or community malaria agents and are either pre-treated with permethrin or permanently treated nets (Permanet). In the case of existing nets, the government offers free re-treatment at selected sites throughout these Zobas. The household survey, described below, was intentionally conducted during the rains to capture the period when members of the respective household were most likely to be using the nets.

Results

A total of 2341 households were visited and a questionnaire was given to a household representative (Table 1). The mean number of ITNs per household was equal to 1.35 across all zobas (1.03 in Debub, 1.50 in Gash-Barka and 1.86 in Anseba). A majority of households (66.8%) across all three zobas reported owning at least one ITN (57.1% in Debub; 70.1% in Gash-Barka and 86.5% in Anseba). The proportion of respondents who report mosquitoes as a cause of malaria transmission was also high (87.4%) across all zobas. Over half of the households said that a case of malaria had occurred (been diagnosed at a clinic) within the past 3 years in Debub (50.2%) and Gash-Barka (65.7%). Nearly a third (29.3%) of all households was within villages with a health clinic.

The proportion of households where it was reported that all children <5 years old slept under an ITN the night before the survey varied considerably by zoba (40.9% in Debub, 63.7% in Gash-Barka and 77.4% in Anseba) (Figure 2). While lower, the proportion of households where all occupants slept under an ITN the previous night followed a similar pattern to that for children (20.9% in Debub, 39.2% in Gash-Barka and 49.8% in Anseba). Overall, over half of all households reported all children <5 years under ITNs the night before the survey, while a third (32.9%) reported all household occupants under ITNs the previous night. Of those households owning at least one ITN, 49.3% reported having all members under the ITN the night before the survey; 82.7% reported having all children under five under an ITN the night before the survey. Over one-third of the households in Gash-Barka (36.8%) and Anseba (43.7%) had an ITN-to-household occupant ratio of at least 1:2, while 23.2% of households in Debub had at least this ratio (Figure 2).

Table 2 presents results of three models. Correct knowledge of malaria transmission was a significant determinant of ITN use and possession among sampled households (Table 2). After adjusting for confounders, households with correct knowledge reported by the respondent were more than twice as likely [adjusted odds ratio (OR) = 2.18, 95% confidence interval (CI) 1.34–3.54] to have all children <5 years old sleeping under ITNs the night before the survey, and over one and a half times more likely (adjusted OR = 1.84, 95% CI 1.21–2.68) to have all household occupants sleeping under ITNs the night before the survey interview, as compared to those without correct knowledge. Households having correct knowledge were also 1.5 times more likely to have an ITN-to-occupant ratio ≥1:2. Households in villages with health clinics, which also serve as ITN distribution centres, were significantly more likely to possess and use ITNs, although this affect was marginally insignificant for ITN use among all household occupants. Zoba was a significant determinant of ITN possession and use, with households in Debub least likely to possess or use ITNs.

Among households reporting ownership of at least 1 ITN, zoba continued to be a significant determinant of ITN use for children <5 and all household occupants, with Debub households least likely to have all children or all occupants under an ITN (Table 3). The total number of ITNs per household was a strong predictor of ITN use: the greater the number of ITNs owned, the more likely it was that they were used by children <5 years and by all occupants. Households reporting a recent history of malaria were significantly less likely (OR = 0.74, 95% CI 0.57–0.96) to have all occupants using ITNs, although this was not a predictor for use by children <5.

Discussion

The purpose of this paper was to examine factors related to the possession and use of ITNs in Eritrea. All evidence suggests that current efforts to disseminate ITNs in Eritrea are working, with a high proportion of households reporting ITN ownership (67%), and ITN-to-occupant ratios inside households moving towards 1:2 in all zobas.

Five main findings emerge from this analysis. First, although ITN ownership is high in Eritrea, not everyone sleeps under an ITN every night. This suggests that although most people have the capacity to protect themselves, not everyone is doing so on a nightly basis, which may increase their overall risk of acquiring malaria parasites. It should be remembered that this survey was conducted just as the rains were tapering off in Eritrea, and therefore net use should have been at its peak. In support of this premise, our data suggests that households with all occupants sleeping under an ITN, as well as households with an ITN-to-occupant ratio of 1:2, were less likely to report a case of malaria within the household. Of course, we do recognize that there is some protective effect from ITNs just through their repellent properties, i.e. their presence in the household (Alaii et al. 2003). Thus, we are not saying that this gap between ownership and use is expected to be entirely closed. This would probably be neither necessary nor possible. At present, our information on low prevalent areas and malaria prevention methods is still too scanty to know what the minimum proportion of the household or children should be under a net during high-risk seasons. However, if one's goal is to ‘break transmission’ (which Eritrea's is), then one can assume that this gap between ownership and use needs to be reasonably small. In addition, the members of some of these households where nets were owned but not used, may in fact have had some protection if the net was hanging and not stored away. Our instrument was not sensitive enough to measure the position of ITNs at night.

Second, correct knowledge of malaria transmission (e.g. mosquito bites are a necessary prerequisite to developing malaria) is a good predictor of ITN use within a household when one looks at all households in the survey (as the denominator). Although other studies investigating the association between bed net use and correct knowledge of malaria transmission failed to find significant associations (Aikins et al. 1993; Cham et al. 1996), and behaviour that is protective of health may occur without any insight into the nature of the relationship between disease aetiology and behaviour (Agyepong & Manderson 1999), it is clear from this study that targeted and specific education campaigns can bolster effective ITN use.

The third finding, however, contradicts the previous (i.e. second) one and begs explanation. Once the household owns a net, the variable measuring correct knowledge of transmission is no longer significant. It is possible that we are witnessing the phenomenon noted in other research related to the adoption and spread of a new technology, whereby people believe it is enough to be an ‘ITN owner’ and accept a free net because they are offered it, rather than because they plan to use it or think they need it. Heat, discomfort, a perception that mosquitoes are not around; all these may reduce the likelihood that a household uses the net available to them every night. A better understanding is needed as to why some people do not use nets, and what factors might improve the correct adoption of this technology once it has been distributed.

Fourth, households located in villages with clinics were more likely to possess and use ITNs, as evidenced by significant associations between proper ITN-to-occupant ratios, all children <5 years old using ITNs and the presence of clinics (Table 2). This finding is consistent with what one would expect, as clinics are used as ITN distribution centres in many areas of the country, which means households in villages with clinics are in closer proximity to free ITNs as compared to those who would have to travel to a clinic to receive a free ITN. Although the presence of a clinic is a good predictor of ITN ownership, not all occupants are using an ITN all the time. This suggests that other efforts may also be needed to disseminate malaria control information and materials to effectively protect all household members.

Fifth, the zoba in which the household was located was predictive of both ITN ownership and use in this analysis. In all three zobas, we observed high ITN ownership and use. Both Anseba and Gash Barka household were more likely to have an ITN-to-occupant ratio of 1:2 as compared to Debub, while in Table 2, where the use is measured relative to all households interviewed, Anseba households were more likely to have occupants under an ITN. However, when use is restricted to only those households that have an ITN, Gash Barka and Anseba switch, and households in the former zoba (which has had a bednet program in existence longer than Anseba) are more likely to be using it. In other words, our results suggest that households in Gash Barka have been exposed to public health messages and discussions about use of ITNs for a longer period of time than those in Anseba, and that as a result Gash Barka households appear to be more willing to use ITNs during peak transmission season.

Because the zoba is most likely measuring unobserved heterogeneity related to landscape ecology, ethnicities, religion, access to information and services and transmission patterns associated with location, it is difficult to isolate the exact reasons why these three zobas are different in terms of ITN ownership and use. However, differences in the timing of ITN distribution, as well as differences in terms of capacity of local program managers to initiate behavioural change, may provide additional explanations for these results.

Conclusion

In conclusion, the NMCP in Eritrea is working well to disseminate ITNs, as evidenced by high ITN ownership. At the global and national levels in Africa much attention is currently focused on increasing access to ITNs, our results suggest that additional efforts may be needed to target areas where the use of ITN is low, especially during peak transmission season. Studies may be needed to disentangle the relationship between the ITN coverage rates and factors that influence ITN use. Although evidence supports the premise that Eritrean households may now be approaching levels of adequate access to ITNs, there remains a significant gap between ownership and use. Reducing the size of this gap must become a focus of public health communicators, social scientists and all health programme managers in malarious areas if the ITN strategy advocated by RBM is to be truly effective. It would be tragic if this strategy were to fail in part because the scientific malaria community assumed, as the family planning community did in the 1950s and 1960s; ‘build it, and they will come’–‘provide/or sell them an ITN and they shall use it.’