Markets, voucher subsidies and free nets combine to achieve high bed net coverage in rural Tanzania

It is estimated that malaria is responsible for 515 million clinical attacks worldwide, 70% of these events are concentrated in Africa [1]. Young African children and pregnant women bear brunt of the burden and at least 18% of childhood mortality on the continent is due to malaria [2]. More encouragingly, the fact that insecticide-treated nets (ITN) prevent malaria has been irrefutably documented [3, 4]. The Roll Back Malaria Partnership and Millennium Development Goals (MDG), therefore, aim to achieve 80% ITN use amongst pregnant women and children below five years of age in Africa, while the US President's Malaria Initiative (PMI) is even more ambitious, aiming for 85% use amongst these same population categories [5‐7]. However, there is growing consensus that this important intervention will only achieve its full potential to prevent malaria if at least one third of the entire population sleeps under ITN, as well as the vast majority of the most vulnerable groups such as pregnant women and young children [8‐12]. This is because residents are protected by not only personal use of ITNs but also by the community-wide effect that their neighbours nets have on mosquito populations. Much as there is increasing call for rapid and sustained achievement of high ITN coverage targeting entire populations [9‐11, 13], including non-vulnerable adults and older children, delivery mechanisms by which this noble goal can be achieved are still actively debated [14‐16]. Until recently, public debate has largely focussed upon the comparative merits of free and market-based strategies for deploying ITNs [15‐18]. While spirited debate over such a potentially important public health issue is welcome [15], it carries a risk that policy makers and donors will perceive a false dichotomy between free and market-based strategies for promoting ITNs. If so, they may overlook an important opportunity to implement complementary strategies for rapidly increasing and maintaining high levels of ITN ownership and use.

Tanzania has been a front-line country for testing the efficacy [19] and effectiveness [20] of ITNs, and has developed a nationwide implementation strategy based on in-country experience [21]. Notably, it was also the first country in which a large-scale cost-sharing scheme for distributing subsidized ITNs was evaluated and shown to improve child survival under programmatic conditions [20]. When Tanzania first decided to take ITNs to scale, mosquito nets were almost exclusively supplied through commercial retailers bundled with insecticide-treatment kits subsidized by the public sector [21]. In 2004, the National Malaria Control Programme (NMCP) introduced a voucher subsidy for pregnant women as part of a nation-wide programme to prevent malaria by enhancing coverage of pregnant women and the young children who share their sleeping spaces during and after the pregnancy. In addition to this national programme, NMCP also assisted a small number of districts including Rufiji on the south-central coast to distribute free bundled nets to under-fives through a child health campaign in mid-2005 with support from partner organizations including UNICEF and the Tanzanian Red Cross [22].

The Interdisciplinary Monitoring Project for Anti-malarial Combination Therapy (IMPACT) had been implementing and evaluating effects on drug resistance of sulphadoxine-pyrimethamine (SP) combined with artesunate (SP+Art) for routine treatment of malaria in Rufiji district southern Tanzania between 2000 – 2006 [23]. Annual household surveys, which included net ownership, use and source, were conducted as a routine part of this study. The coincidence of the unsubsidized market, voucher subsidies and free distribution happening at the same time and place created an opportunity to evaluate the interactions between these major and apparently inconsistent ITN delivery strategies – the primary focus of the ongoing debates. This paper presents results from this assessment and show that these combined strategies complemented rather than competed with each other.

Net use varied across population age groups (Pearson χ² (d.f. = 12) = 839.9253; P < 0.001) with excellent targeting of high coverage to the most vulnerable groups (Table 1). Infants (0–12 months) had the highest proportion of net use in the study area, with >85% using any net the previous night, exceeding the targets of the Millennium Development Goals (MDG), Roll Back Malaria (RBM) and the US President's Malaria Initiative (PMI) for ITN use [5‐7]. It is true that these targets have only been exceeded in terms of any net and that approximately half of these were not recently treated. However, long-lasting insecticide formulations [29, 30] for factory pre-treatment or bundling with all nets made in Tanzania were introduced as of March 2007 so these documented levels of coverage with any net should practically translate such achievements into de facto ITN coverage. Coverage of young children exceeded the MDG and RBM targets but not the PMI target for protection of under-fives with usage rates exceeding 80% [5‐7]. While coverage of adults and older children was lower than that of the vulnerable groups that were targeted with the bulk of the subsidies, overall coverage of the entire population as a whole was more than sufficient to achieve major communal reduction of malaria transmission [9, 10] if new long lasting treatments could make most nets insecticidal for their lifetime. Although only short-lived insecticide formulations were available at the time, approximately one third of the population used a recently treated net in 2006 so appreciable communal suppression is likely to have been achieved [10, 31]. It is expected that such invaluable community-level benefits to be improved upon by the superior ITN technologies which are now available [31, 32] and further gains in coverage as these delivery systems become better established.

It is noteworthy that proportional contribution of various net delivery systems varied across age groups (Pearson χ² (d.f. = 20) = 844.8122; P < 0.001) and each delivery system appears to have supported its appropriate target group (Table 2). The majority of nets used by infants were obtained through the national voucher scheme, indicating this important vulnerable group is effectively targeted by this system for delivering heavily subsidized nets through commercial distributors. Nets provided at no charge to the user during the child vaccination campaign supported half of the net coverage achieved amongst young children and over a third of coverage amongst older children. The commercial market, with no subsidy other than bundled insecticide, was the biggest source of nets in the population as a whole. Nets obtained at full market price accounted for almost two thirds of use by older children and adults who must be covered if community-level suppression of transmission is to be achieved [10]. Clearly this mix of delivery mechanisms reaching different segments of the population demonstrates that all three delivery tactics are complementary rather than competitive. Rapid attainment of high net coverage for the vulnerable population was achieved through the combined contributions of the product provision campaign and voucher subsidy while broad coverage for the rest of the community resulted largely from nets purchased on the open market at full price [21]. Viewed in this integrated manner, these data provide clear evidence that commercial markets, voucher subsidies and free net distribution are not mutually exclusive choices and can complement each other effectively to make the most of limited subsidy.

Figure 2 shows the number of nets in use during the 2006 survey by source and time of acquisition. The distribution of nets at no cost to the recipients through the vaccination campaign in the third quarter of 2005 caused a clear surge in net acquisition. It is interesting to note that a concomitant surge was observed for nets obtained through sundry other sources, suggesting significant redistribution within families and communities. Over 16% of all nets reported for this period were obtained through this mechanism. It therefore seems likely that under-five children who already had a net and received another through free distribution from the campaign may have passed on the existing or additional net. This suggests that every additional net supplied contributes to both personal and communal protection in the population as a whole, regardless of any leakage or exchange. While the number of nets procured on the unsubsidized market or through the voucher for pregnant women declined shortly after the free distribution in mid-2005, within a year these market-based sources of nets appear to have rebounded to levels equal to or in excess of their levels before the campaign. Provision of nets at no cost through the public sector did not compromise the viability of either the voucher scheme or the commercial market, presumably because this limited full subsidy was targeted toward a previously unsubsidized population group and was not of sufficient volume to compete with established demand for nets in the entire population. Such hybrid approaches represent an excellent mix of strategies for catching up and keeping up coverage when subsidies are not adequate for mass distribution to entire populations [32, 33]. This suggests that a number of options are available to NMCPs and that diverse tactics can be astutely combined to achieve the RBM, MDG and PMI targets rapidly and sustainably, even with only partial subsidies.

Table 3 demonstrates that contributions of various delivery mechanisms to net use vary by socio-economic status (Pearson χ² (d.f. = 20) = 844.8122; P < 0.001). Net coverage was far higher for the least poor with more than four fifths of this better off quintile using a net. Over half obtained their net from commercial market at full price. The concentration index of inequality was the highest for nets obtained from this almost completely unsubsidized distribution mechanism (Table 3) and the concentration curve for these nets lies below the line of perfect equity (Figure 3). This confirms favouritism of the unsubsidized market towards those who are better off and can readily pay for nets. Nevertheless, unsubsidized commercially sold nets were the most important for net coverage achieved by all other socio-economic groups except for the poorest where this source was matched but not exceeded by those obtained for free from the child vaccination campaign. Indeed, the contribution of the unsubsidized commercially-obtained nets towards net coverage achieved by the poor exceeded that of nets obtained with assistance of the voucher subsidy, demonstrating that even the poorest invested in nets for non-target groups when no subsidy was available (Table 2).

Conventional interpretation of concentration indices, comparing coverage of the poorest with that of the least poor, might suggest that all forms of subsidized delivery (voucher and free product provision) are completely equitable (Table 3). Similarly, uncritical interpretation of Figure 3 might confirm this observation with all these forms of subsidized delivery approximating the line of perfect equity. Closer examination of Table 3, however, indicates that the least poor may not be a representative group with which to compare the poorest because this group appears to underutilize subsidized delivery mechanisms and relies more heavily upon the open market, presumably for reasons of choice and convenience. Comparing the poorest with the three intermediate wealth quintiles shows that the former do suffer substantial inequities relative to the latter, regardless of what system is used to deliver subsidized nets. While coverage of the three intermediate wealth quintiles is relatively even, coverage of the poorest is consistently and substantially lower for both voucher-subsidized and freely distributed nets. Nevertheless, population-wide net coverage for the most poor approaches the levels at which community level protection may be achieved [10] when the reported long lasting net treatment technologies[30] will make all the nets insecticidal. It also should be noted that the poorest are typically intermingled with neighbours from better socioeconomic strata and, therefore, share the communal protection delivered by the coverage of these groups which is typically twice as high.

This study has reported how commercial markets, free product distribution through mass campaign and voucher subsidy can work together to achieve high rapid and sustainable ITN coverage in a poor rural African community exposed to intense and perennial malaria transmission. Some recent publications have advocated that ITN coverage targets can only be achieved through mass distribution at no cost to the end user [15‐17]. While these arguments are based largely on opinion, this study provides empirical data showing that direct product distribution through mass campaigns can accelerate attainment of coverage in target groups if astutely integrated with market based approaches which similarly target subsides to the neediest. Vouchers and free products both clearly succeeded in targeting their respective subsidies to different biologically vulnerable population groups they were intended to support (Table 2) with insecticide treatment rates being highest in the vulnerable groups (Table 1). Not only did both approaches bias coverage of nets and insecticide treatment to the young (Table 2), in combination they enabled a largely unsubsidized private sector to flourish, resulting in high coverage of the non-vulnerable majority of the population. This latter point is crucially important for achieving community-wide suppression of transmission [9, 10, 12] and represents a step forward relative to recent studies in Kenya and Ghana which focused exclusively on coverage of infants and young children [32, 34]. When data for infants and young children are pooled, use of any net for under-fives observed in this study is higher 1020/1216 (83.9%) than that reported from Kenya (80.3%) [34] and in Ghana (72.6%) [32]. Use of recently treated nets shown amongst under-fives appears to be lower in this study (43.4%; 528/1216) than that achieved in Kenya 67.3% and Ghana 59.6%. Nevertheless, it is suggested that the issue of net treatment rates will become less challenging as NMCPs increasingly prioritize the exclusive promotion of long-lasting insecticidal nets and treatments [12].

Importantly, largely unsubsidized nets were available to everyone able and willing to pay for them through the commercial market which was actively promoted through the voucher scheme [21]. Indeed, the majority of all nets used in our study area were obtained at full market price reflecting the contribution of the community itself to the cost of high population-wide coverage under current circumstances in which global public subsidies for malaria control amount to only 20% of the true full cost [35]. Although use of unsubsidized nets was greater amongst the least poor (Table 3), recently voiced concerns about the potential inequities associated with market-based delivery of subsidies [16, 34] appear to apply just as much to free product delivery, with approximately equivalent inequity resulting from both voucher discounts and fully subsidized distribution of nets through vaccination campaigns.

Our study has shown that overall net use in Rufiji district was far higher than most other parts of Africa. This data provides further definitive evidence that net delivery strategies other than fully subsidized mass distribution to entire populations can achieve net coverage high enough to provide community level benefits. Achieving high ITN coverage for non-pregnant adults and older children is just as important as comprehensive personal protection of vulnerable groups for three reasons. First, they are the majority in the population and more attractive to mosquitoes [36‐38] so reasonably high but not necessarily comprehensive net coverage is essential to deliver the mass effect of ITNs [9, 10, 12]. Second, they are the only source of labour for economic productivity required to support the population as a whole and dependent children in particular, so the impacts of malaria illness and associated costs trickle down to every one [39]. Third, many people living with HIV may be biologically vulnerable to severe malaria in a manner similarly to small children and pregnant women [40, 41]. Here it is demonstrated that high net coverage for adults and older children has been achieved in our study area, largely through purchase of unsubsidized nets at market prices. This study supports the view that high and broad ITN coverage including older children and adults is important for effective malaria control [9‐11] but show here that market-based cost-sharing strategies utilizing voucher-targeted subsidies can also help achieve this goal [16]. However, unlike previous reports from other parts of Tanzania [14], this is evidence that such targets can be achieved very rapidly by augmenting voucher-stimulated "keep-up" mechanisms with complementary "catch up" campaigns directly distributing products at no cost to the end user.

One limitation of this study is that our survey did not distinguish between pregnant and non-pregnant adult females so coverage in this key target group could not be directly assessed. Nevertheless, high net usage by infants, largely supported by the voucher scheme, presents an informative proxy for net use by pregnant women because mothers of newborns tend to share sleeping sites with their offspring in this area. All in all, Tanzania may be a uniquely informative site for a study of this kind. It was the site of initial experimental hut trials and ITN efficacy studies [19] and it is where such innovations as do-it-yourself treatment and social marketing were pioneered [42]. Moreover, the combination of both a socialist past and more recent reforms to enable a market-based economy may have elevated many of the national and local policy makers above the ideological considerations that have too often characterized discussions about how to deliver ITNs. The open market and subsidized voucher programme were relatively mature at the time the free distribution was undertaken in Rufiji District and this may not be a unique situation. Market-based delivery systems for ITNs are operational in many countries and present a valid option for attaining high coverage without comprehensive subsidies. As donor support for mass distribution of free nets become more widely available, and it is sincerely hoped that this will be viewed as potentially complementary rather than disruptive to market-based distribution and promoted as a means for rapidly expanding coverage, particularly amongst vulnerable groups. Unlike the development and implementation of the voucher programme [21], the decision to undertake a mass distribution in Rufiji District was not achieved by broad national consensus and was more opportunistic in its origins. Instead, the mix of strategies employed in Rufiji District represents the pragmatic efforts of local and national health officials to sensibly deploy scarce resources offered by partners with competing ideologies. In that sense, it is expected that Tanzania and Rufiji District will not remain an historic exception. If, decades after the life-saving value of this intervention has been firmly established [3], ITN advocates, no matter their stripe, can agree that broad community coverage is a priority, then there is promise that diverse approaches can be simultaneously applied in an imperfect but constructive and complementary manner. It is time to step away from the ideological debates about how to deliver ITNs and engage constructively with the local and national authorities to confronting this challenge through whatever means are practical.