Monitoring, characterization and control of chronic, symptomatic malaria infections in rural Zambia through monthly household visits by paid community health workers

A relatively large proportion of malaria infections are only mildly symptomatic, especially in endemic countries where acquired immunity moderates both parasitaemia and pathology [1‐3]. Even in areas with only modest, seasonally sporadic transmission where little immunity exists among the human populations, the many mildly symptomatic infections that persist chronically, often at sub-patent parasite densities below thresholds of detectability, are responsible for sustained malaria transmission [4, 5]. As early as the 1930s, these populations have been targeted for treatment with anti-malarial drugs as a control measure for malaria transmission, particularly during the era of the Global Malaria Eradication Programme initiated in the 1950s [6‐8]. Due to the emergence of drug resistance, which was attributed to mass drug administration campaigns [9‐11] and evidence that the impact on transmission may be limited, the use of therapeutic approaches to control transmission rather than burden was not considered as effective [6, 8, 12, 13] until recently, when interest in malaria elimination was rejuvenated [13‐15].

Currently curative drugs are predominantly targeted at symptomatic individuals based on passive detection of acute infections among patients seeking care for fevers through the formal and informal health system [16‐18]. Strategies for control of human-to-mosquito transmission by providing chemotherapy to chronic parasite carriers are now being revisited as they may be complimentary to front-line interventions for preventing mosquito-to-human transmission, such as long-lasting insecticidal nets (LLINs) and indoor residual spraying (IRS) of houses, to further accelerate declines in malaria burden [19‐21] in areas of low and seasonal transmission [8, 13, 22].

However, most of the infected fraction of the human population only exhibits sub-acute symptoms so they do not always obtain medical care, even where it is readily available [23, 24]. These carriers of chronic, mildly symptomatic malaria infection thus act as the silent reservoir of infection because even low, often sub-patent levels of parasitaemia are sufficient to infect mosquitoes [25‐27]. Additionally, asymptomatic individuals risk developing chronic anaemia and symptomatic malaria [28, 29]. Across all levels of transmission, the majority of infections and onward transmission to mosquitoes occurs in older age groups, even though the young often harbour the highest parasite densities, because the former comprise the bulk of the human population and vector biting rates increase with host biomass and therefore age [20, 30, 31]. In such populations, many individuals remain parasitaemic and infectious for one or more years [32‐34]. Thus, in order to eliminate human-to-mosquito transmission with therapeutic drugs, all cryptic or asymptomatic infections within a human population must be successfully terminated, necessitating comprehensive coverage of targeted communities [20, 21, 28, 35].

Essentially two broad strategies for taking malaria chemotherapy beyond routine case management have been described: mass drug administration (MDA) or mass screening & treatment (MSAT), historically referred to as mass blood examination [20, 36]. MDA entails administering anti-malarials to every traceable consenting member of a population, regardless of whether their malaria infection status is known or whether they exhibit symptoms [13, 20, 36], while MSAT targets only confirmed parasitaemic individuals after parasitological testing for infection status [20, 36]. MDA necessitates comprehensive coverage and failure may accelerate the spread of drug-resistant strains selected by strong but incomplete selection pressure on the parasite populations [37, 38]. MSAT has been proposed as a better alternative because treatment is limited to those diagnostically confirmed to be infected, thus lowering treatment costs and risks of selection for resistance [39, 40]. However, the major limitation of MSAT lies in the challenge of detecting low-density infections, which contribute substantially to the reservoir of infection [40]. With increasing availability of more sensitive, scalable malaria rapid diagnostic tests (RDTs) that can be used by non-specialist community based health workers in low-resource settings, MSAT is now logistically feasible [38, 39, 41, 42].

The concept of community-based management of malaria stems from the recognition that human resource deficits amongst clinically-trained professional cadres are commonplace, so extending service delivery beyond centralized health facilities, by mobilizing through community health workers (CHW) will be required to improve access to appropriate management of uncomplicated malaria [43, 44]. CHWs have demonstrated the capacity to effectively diagnose malaria with RDTs and provide treatment according to the locally relevant policy and guidelines [41, 42, 45, 46]. The community-based diagnosis and treatment approach has also been shown to be cost-effective [47, 48], improve delivery of malaria case management overall [49‐51], is well accepted by communities [52‐54] and also provides a potentially valuable population–wide platform for monitoring trends in human parasitaemia [55].

However, this approach remains grossly underutilized and understudied, with only 15 million RDTs utilized at community level globally, mostly in India [56]. This study therefore evaluated the effectiveness of paid CHWs providing improved access to blood screening and treatment services to community members, not only when they self-reported because they felt ill, but also through regular monthly active visits to their homes.

Despite the fact that only a quarter of the covered resident population agreed to be tested in each monthly round of household visits by the CHWs, these active surveys by modestly remunerated paid CB staff identified >11,000 malaria parasite infections in a population of <18,000 residents in a single calendar year, of whom more than half were symptomatic in or around the time they were visited and may not have otherwise sought care. The far higher sensitivity with which these active household surveys by the CHWs detect cases of malaria infection is also reflected in the observation that this surveillance arm captured twice as high an incidence rate as the passive surveillance activities of the HFs and CHWs combined. The strong association of many symptoms, especially fever, headache and vomiting with malaria infection, particularly among individuals tested during active household visits by the CHWs, confirms previous reports that illustrate just how inaccurate the term asymptomatic is in relation to widespread chronic malaria infections [79‐81] that clearly cause very large proportions of the overall burden of clinical illness in the community (Table 1, Figure 7). Clearly a large proportion of the population are infected with malaria, and suffering from a range of mild symptoms of clinical illness as a consequence, but do not feel ill enough to report to a HF or even to a nearby CHW to seek care. In addition to representing a major proportion of overall morbidity burden among the population, these chronic infections also act as a reservoir for continued transmission [4, 5, 27, 80, 82]. Regularly scheduled household visits by CHWs, who presumably will need to be paid for such a full time commitment, may therefore be extremely useful for identifying, treating and mapping the individuals who harbour chronic malaria infections and constitute the infectious reservoir that sustain transmission [38, 83, 84].

However, as implemented in this study, even these monthly active household visits, repeated on a continuous monthly survey cycle, had no obvious impact on malaria infection burden, possibly because most participants did not consent to testing often enough to benefit from any impact upon malaria infection and associated symptoms that are suggested, but not proven by, Figures 4 and 7, respectively. While the AL treatment used here has well-documented gametocidal properties [85‐88], the limited sensitivity of RDTs or microscopy and considerable natural density fluctuations of circulating P. falciparum blood stages mean that approximately half of all malaria infections are sub-patent and escape detection by a single testing event [27, 89, 90]. Furthermore, mosquito-to-human transmission remained remarkably high in the study area, measured as a mean EIR of approximately 70 infectious bites per unprotected person per year [57, 58]. The most likely explanation of the persistence of such intense transmission, despite reasonably high rates of LLIN use, supplemented with IRS in selected clusters towards the end of this study, is probably the emergence of pyrethroid resistance among local populations of An. funestus[91]. The high rates of re-infection suggested by these entomological surveys are consistent with and confirmed by the high rates of infection incidence [92] recorded here (Table 3) despite the imperfect sensitivity of RDTs [93, 94]. Given the imperfect detection sensitivity of RDTs [93, 94] and the rapid rates of re-infection that can be expected in a setting with such a high EIR [30, 95‐98], it is unsurprising that at least monthly screening and treatment is required to achieve dramatic reductions of malaria infection burden (Figure 4B), associated symptoms (Figure 7), and presumably transmission [27, 99, 100], even assuming these two figures reflect genuine impact rather than mere association. However, it is certainly encouraging that the apparent impacts among residents consenting to such frequent testing and treatment, which these associations suggest, compare very well with simulations and field data from annual mass screen and treat programmes [40], and even simulations and field observations of year-long mass drug administration programmes with treatment cycles of only four or even two weeks [19]. It may also be encouraging that, despite their known limited sensitivity, RDTs appear to be sensitive enough to detect persistent malaria infections if each individual is tested often enough (Figure 4B) so that the frequent sporadic surges of detectable parasitaemia characteristic of P. falciparum are captured [101, 102]. If the observed association of parasitaemia with testing and treatment (Figure 4B) reflects genuine impact, this also suggests patient compliance with the AL treatment regime used in this study was probably comparable with high estimates (84.5%) from previous evaluations in Zambia [103].

Beyond extending delivery of diagnostic and therapeutic services to the grass roots community level, the CHWs also provided a remarkably informative source of surveillance data, including the overall burden and distribution of malaria and associated clinical symptoms, a number of important demographic and geographic determinants of risk, and the rates utilization of preventive interventions, such as IRS and ITNs. It is particularly encouraging that, despite their known sensitivity limitations, RDTs [93, 94] appear to be more than adequate for monitoring disease burden through CHW extension systems that can guide programme implementation. Latent antigenaemia several weeks after successful clearance of infection can cause false positive results when using HRP2 based RDTs, and therefore over-prescription of anti-malarial drugs [104‐106], so it is possible that estimates of cost and cost-effectiveness described in Table 3 may be improved upon with better diagnostic technology. If scale up of such CHWs is affordable beyond this research setting and could be scaled up across entire districts, provinces or even whole countries, such routinely collected data reported in disaggregated form from such small population subdivisions could be invaluable at all levels of programmatic monitoring and evaluation.

The costs of providing this CB extension of primary health care services, to provide both active and passive screening and treatment for malaria were substantive (Table 3), corresponding to 11.1% of the annual per capita health budget of Zambia in 2011 ($96) [107]. Furthermore, to achieve the full potential of this service by ensuring community-wide engagement in screening and treatment on at least a monthly basis (Figure 4B), these costs are likely to double, even if baseline levels of transmission were reduced to pre-elimination levels so that the costs of drug treatment were negated (Table 3). It is highly unlikely, or desirable, that such a cadre of CHWs would be mobilized to deal with surveillance and control of malaria alone so these CB personnel would also be required to deal with uncomplicated forms of other common illnesses like diarrhoea and pneumonia [51, 54, 108]. Even the passively provided malaria diagnosis and treatment services described here would need to be augmented with a range of other clinical services to be supported at programmatic level [51, 54]. It is therefore difficult to envisage CHWs effectively or sustainably taking on such substantive commitments on a purely voluntary basis with no remuneration whatsoever. Barely more than a third of the overall costs of the active household surveys by the CHWs were accounted for by the cost of their meagre remuneration, and this figure would reduce to less than a fifth if the increased commodity costs of full compliance with monthly screening and treatment were to be incurred through improved community participation. Paying these CHWs is therefore not only likely to be essential to ensure their retention and effectiveness as full time agents of malaria infection surveillance and control, it is also a relatively minor fraction of the overall cost of actively delivering extended CB primary healthcare services.

Apart from its observational design, the most obvious limitation of this study is that the majority of enrolled residents were tested and, where appropriate, treated far less than once a month (Additional file 1, Figures 2A and C, 3C and D, and 4A) as originally envisaged. Future evaluations of CHWs, especially those engaged to conduct frequent active household surveys of their entire assigned populations, should include operational research studies to better understand and address the limitations of service uptake by community members observed in this study.

The monthly active household visits to entire communities by CHWs equipped with existing field-compatible diagnostic tools were not sufficient to eliminate the human reservoir of malaria infection from this rural African setting with intense transmission despite reasonably high LLIN/IRS coverage. However, observed negative associations between infection status and frequency of testing and treatment suggest that dramatic impact upon malaria parasite infection risk and associated disease burden may be achievable through far more regular testing and treatment. Substantive alleviation of malaria may be attainable and cost-effective if the substantial, but not necessarily prohibitive, costs of implementing frequent active CB surveys for chronic malaria infections are affordable to national programmes and higher levels of community participation in regular testing opportunities are achievable.