Detection of foci of residual malaria transmission through reactive case detection in Ethiopia

The global pattern of malaria epidemiology has changed remarkably over the last decade. Between 2010 and 2015, malaria mortality rates declined by 35% among children under 5 years of age, with 21% fall in incidence among the population at risk [1]. Implementation of key malaria prevention and control measures have played a pivotal role in decreasing morbidity and mortality due to malaria [2, 3]. Furthermore, concerted control and elimination efforts of malaria in some countries, such as the United Arab Emirates and Sri Lanka, have resulted in malaria-free status in recent years [4, 5]. Despite the achievements gained in the control of malaria, the disease still remains a significant public health problem in many sub-Saharan African countries, including Ethiopia.

The epidemiology of malaria in Ethiopia appears unique, compared to other countries in sub-Saharan Africa, in that both Plasmodium falciparum and Plasmodium vivax coexist. While almost all cases of malaria are due to the two species, there is high spatiotemporal heterogeneity in the distribution of these parasite species. According to the 2015 National Health Sector Development Plan report [6], out of the total microscopy or rapid diagnostic test (RDT) confirmed malaria cases, 63.7% and 36.3% were due to P. falciparum and P. vivax, respectively. In a recent study done in Jimma in south-western Ethiopia, however, more than three-quarters of the cases were due to P. vivax [7]. Plasmodium ovale plays a minor role in Ethiopia, and appears to be often misdiagnosed [8].

Over the last decade, during which malaria elimination was put back on the global health agenda, morbidity and mortality due to malaria has remarkably declined in Ethiopia [9, 10]. Besides the sharp decline of malaria including from some of the historically malarious areas of the country [11], no major malaria epidemics, which usually recur every 5- to 8 years, have been reported since 2005 [12]. Implementation and scale-up of the powerful vector control interventions, including indoor residual spraying (IRS) and long-lasting insecticidal nets (LLINs) appear to have played key roles [13]. More than 17 million LLINs have been distributed in 2014/2015 alone, with cumulative number of the nets distributed since 2009 being scaled up to more than 75 million [6]. Access to malaria diagnostics and treatment has also remarkably improved over the last decade, mainly via the innovative health extension programme [14] that operates at community level.

Based on the malaria control achievements gained, and with the help of international partners, Ethiopia has set goals to eliminate malaria by 2030. However, substantial portions of human Plasmodium infections are asymptomatic, often remaining undetected by microscopic examination [15]. Asymptomatic infections can serve as reservoirs of infection to the vector mosquitoes [16], potentially sustaining transmission.

To further sustain control of malaria and move towards elimination, adequate detection and prompt treatment of both symptomatic and asymptomatic cases in the community is critical [17]. One of the strategies of addressing malaria cases not presenting to the health care facilities is reactive case detection (RACD) with focal test and treatment methods. Reactive case detection makes use of the spatial clustering trend of malaria carriers particularly in low endemic settings [18, 19]. Hence, in RACD, following passive case detection, household members of the index case and neighbours located at certain distance from the index household are screened. This method has been utilized in several low malaria transmission settings [20, 21], despite lack of established standard approach to the spatial range of neighbouring households to be within the screening radius.

Reactive case detection also allows detection of asymptomatic malaria infections, which play a major role in sustaining malaria transmission in low-transmission settings [22]. However, active case detection of malaria is not yet fully implemented in the routine health care system in Ethiopia. Thus, this study is aimed at detecting malaria cases using RACD in two health centres in Jimma Zone, south-western Ethiopia.

This study revealed that index cases presenting to health facilities enabled detection of good number of additional asymptomatic malaria in the community. Thus, screening of individuals using PCR following index malaria cases presenting to health facilities is a valuable method for detecting additional malaria cases in low transmission settings. However, it should be noted that PCR-based malaria detection techniques are largely restricted to research settings, and not currently in use by the National Malaria Control Programme in Ethiopia. This calls for evaluation of affordable alternative test methods with better sensitivity compared to microscopy to be utilised in elimination settings.

The overall prevalence of malaria detected using the RACD was 8.96%, more than 90% of the cases being due to P. falciparum. This is in contrast to the recent report from Jimma, in which less than a third of the malaria cases were due to P. falciparum [7], and the national report (63.7% of the documented malaria cases being caused by P. falciparum) [6]. There might be high spatio-temporal variation of Plasmodium species in a country, and even among villages within the same district, mainly due to difference in ecological, climatic and socio-economic factors, and interventions carried out to control malaria [28, 29]. Significant clustering of Plasmodium infection between kebeles, and households was observed in this study. The vast majority of cases were found in one of the eight kebeles included in this study, and individuals residing within the index house were at twofold higher risk of Plasmodium infections as compared to their neighbours. This has an important implication in prioritizing resources for targeted malaria control and elimination efforts in low transmission setting.

Apart from household-level clustering of the malaria cases, significantly higher prevalence of malaria was documented in Doyo Yaya kebele compared to other kebeles. Heterogeneity in transmission of malaria is a well-known feature [30]. It appears that the clustering of malaria in this kebele is likely due to presence of a village not covered in the preceding IRS operation. Indeed, most of the houses which did not receive IRS within the preceding 12 months of the survey were located in this kebele. Malaria control interventions in resource-limited areas should, therefore, target such hotspots that may contribute to sustained transmission, and possibly fuel epidemics in such settings. It is worth noting that screening of malaria was limited to approximately 200 m from index houses in this study, and thus the overall prevalence in the population was not known. Several studies utilized different screening radii around the index houses, often detecting more cases within the index household and among those residing closer to the index houses [20, 31]. However, the optimum screening radius from the index house that should be included during RACD still remains unclear. It appears that the screening radius is mainly influenced by the local epidemiology of malaria and available resources to implement RACD [32].

Improved housing structure is essential in deterring endophagic/endophilic mosquitoes from reaching the occupants [33], thus possibly reducing vector-human contact. In this study, individuals residing in houses with open eaves were significantly more likely to succumb to malaria compared to those living in houses with closed eaves. Significantly higher odds of malaria in houses with open eaves was also reported previously [34, 35]. Earlier reports around the study area indicate that Anopheles gambiae sensu lato (presumably Anopheles arabiensis), the primary malaria vector in Ethiopia, to be predominantly endophagic [36], which may explain the observed higher prevalence of malaria in those living in houses with open eaves. Significantly higher number of indoor-resting malaria vectors were also observed in houses with open eaves in a previous study from central Ethiopia [37].

The vast majority of the malaria cases detected in this study were asymptomatic, most of which were sub-microscopic. Reports also show that in low-transmission settings, asymptomatic infections are common and most of the asymptomatic infections are sub-microscopic [22, 38]. As asymptomatic malaria carriers apparently do not seek treatment, they may serve as reservoirs of infection [39], jeopardizing elimination efforts. The challenges asymptomatic malaria infections pose to the malaria elimination efforts is also exacerbated by the diagnostic limitations in detecting the infections [40]. In very low transmission settings, sub-microscopic carriers may contribute up to 50% of human-to-mosquito transmissions [41]. Apart from the possible contribution to onward transmission, persistent asymptomatic infections may be associated with other deleterious health outcomes, such as anaemia [42].

In the current study, it was also found that self-reported history of malaria in the preceding 1 year before the survey was significantly associated with Plasmodium infection, after adjusting for the other variables. Individuals with history of malaria were more than three times likely to have Plasmodium infection. Relapse with vivax malaria is well known, hence, vivax cases could be due to relapse, even if small number of P. vivax cases were detected. It could also be attributed to persistent P. falciparum asymptomatic infections. The duration of persistence of asymptomatic P. falciparum parasitaemia is debatable. A recent report from Cambodia shows limited duration of asymptomatic P. falciparum parasitaemia in low transmission setting [43], while cases of asymptomatic falciparum malaria persisting more than a decade have been documented elsewhere [44].

In this study, PCR detected 2.2-fold higher Plasmodium infection as compared to microscopy. This difference is comparable with other previous studies [22, 45]. As the current routine malaria diagnostic protocol in health facilities in Ethiopia involve blood film microscopy, it is inevitable that sub-microscopic infections remain a huge challenge to the envisaged malaria elimination efforts. Thus, alternative field-friendly and more sensitive diagnostic tools such as highly sensitive RDTs (HS-RDTs) or loop-mediated isothermal amplification (LAMP) need to be evaluated and used for detection of sub-microscopic/asymptomatic malaria.

The difference in the prevalence of malaria was not significantly different among different age groups and between sexes. However, family size was a one of the main predictors. In households with five or more occupants, the risk of infection was threefold, compared to those living in families with fewer than five family members. This could be due to higher number of anopheline mosquitoes related with increased number of household occupants [46].

Long-lasting insecticidal nets were the only personal protection tools utilized by the study participants. However, the difference in prevalence of malaria among households with sufficient number of LLINs was not significant. Similarly, there was no significant difference in LLIN usage the preceding night on malaria risk. These could perhaps be due to poor integrity of the LLINs, rendering only partial protection from mosquito bite, and/or possibly inaccurate information provided on the number of nets owned and their usage. The apparent lack of association of sufficient ownership or usage of the LLINs and Plasmodium infection could also be related with biting activity of the anopheline mosquitoes in the area. It was reported in a neighbouring district that the peak biting time of An. gambiae s.l. was at the early part of the night, before 21:00 h [47], during which most of the inhabitants likely do not sleep. As the LLIN coverage was high in this study, it may have ‘herd effect’, in that, those not utilizing the nets may be protected as a result of usage of the rest [48].

Sustained residual transmission of malaria in the areas necessitates assessing vector behavioural characteristics and human activities that may contribute to the on-going transmission. As LLINs and IRS interventions target vectors which feed and rest indoors, malaria elimination using these interventions alone may not be achieved [49]. Outdoor biting by the vectors as a result of long-term use of the vector control tools, and resistance to insecticides used in IRS and treating the LLINs may contribute to sustained low transmission of malaria [50, 51]. Moreover, night-time human activities may increase malaria risk [52].

The study has the following limitations: First, due to limitation of resources, the RACD was limited to catchment kebeles of the two health centres. Hence, passively detected malaria cases which resided out of the catchment of the two selected health centres were not included. Second, this study did not include index cases who might have visited private clinics. However, the areas being predominantly rural, it is expected that most of the residents visit the two public health facilities for treatment of fever. Finally, while this study allowed to identify high number of Plasmodium infections around the index cases from one site, no individuals living > 200 m from index case houses were screened for infection. Thus, it was not known how large this cluster of transmission was.

The study revealed that substantial number of malaria cases, largely asymptomatic, were detected using RACD. Reactive case detection strategies utilizing malaria diagnostic tools of high sensitivity, therefore, complement the routine passive case detection in low malaria transmission areas, and may be used in malaria elimination programme. Significant household and kebele level clustering of the malaria cases were observed. Moreover, history of malaria, houses with eaves and family size were risk factors to Plasmodium infection. Further studies incorporating greater distance from index households, and including environmental factors affecting risk of malaria infection are recommended.