Reactive Case Detection Strategy for Malaria Control and Elimination: A 12 Year Systematic Review and Meta-Analysis from 25 Malaria-Endemic Countries

Reactive case detection (RACD) is the screening of household members and neighbors of index cases reported in passive surveillance. This strategy seeks asymptomatic infections and provides treatment to break transmission without testing or treating the entire population. This review discusses and highlights RACD as a recommended strategy for the detection and elimination of asymptomatic malaria as it pertains in different countries. Relevant studies published between January 2010 and September 2022 were identified mainly through PubMed and Google Scholar. Search terms included “malaria and reactive case detection”, “contact tracing”, “focal screening”, “case investigation”, “focal screen and treat”. MedCalc Software was used for data analysis, and the findings from the pooled studies were analyzed using a fixed-effect model. Summary outcomes were then presented using forest plots and tables. Fifty-four (54) studies were systematically reviewed. Of these studies, 7 met the eligibility criteria based on risk of malaria infection in individuals living with an index case < 5 years old, 13 met the eligibility criteria based on risk of malaria infection in an index case household member compared with a neighbor of an index case, and 29 met the eligibility criteria based on risk of malaria infection in individuals living with index cases, and were included in the meta-analysis. Individuals living in index case households with an average risk of 2.576 (2.540–2.612) were more at risk of malaria infection and showed pooled results of high variation heterogeneity chi-square = 235.600, (p < 0.0001) I2 = 98.88 [97.87–99.89]. The pooled results showed that neighbors of index cases were 0.352 [0.301–0.412] times more likely to have a malaria infection relative to index case household members, and this result was statistically significant (p < 0.001). The identification and treatment of infectious reservoirs is critical to successful malaria elimination. Evidence to support the clustering of infections in neighborhoods, which necessitates the inclusion of neighboring households as part of the RACD strategy, was presented in this review.


Introduction
Malaria, a vector-borne disease, remains a major public health problem. Globally, 84 countries with malaria endemicity recorded an estimated 247 million cases of the disease in 2021, up from 245 million in 2020, with most of this rise occurring in countries in the World Health Organization African Region [1]. However, the malaria elimination effort is gathering momentum in many countries. The total number of malaria-endemic countries that counted fewer than 100 malaria cases increased from 6 in 2000 to 27 in 2021 [1]. During the same period, the number of countries that recorded fewer than 10 indigenous cases rose from 4 to 25 [1].
Over the years, milestones such as these have hinged on vector-control strategies such as periodic indoor residual spraying (IRS), distribution of long-lasting insecticide nets (LLINs), increased funding, strengthening of health systems, seasonal malaria chemoprevention in children, preventive chemotherapies (e.g., intermittent preventive treatment in infants and pregnant women), and improved case reporting and surveillance [2]. Today, the recommendation by the WHO on the use of the RTS, S malaria vaccine for the prevention of Plasmodium falciparum malaria in children living in moderate-to high-transmission areas, as described by the WHO [1], will further complement these existing interventions. Some countries focus their malaria prevention strategies on the above-mentioned interventions together with malaria control programs that aim at a decline in the disease burden until it ceases to be a public health problem. Other countries with fewer than 10,000 malaria cases work towards elimination to guarantee sustained zero local transmission of malaria in the population within a specified geographic area through enhanced surveillance systems.
Asymptomatic malaria infections have existed for many years, and finding and treating individual asymptomatic infections, which comprise 60% of the infected population, remains a challenge [3]. An infected asymptomatic individual can become an important reservoir of transmission to healthy individuals under suitable conditions [4] and present an obstacle to the elimination of the disease. In hypoendemic areas moving towards malaria elimination, asymptomatic malaria must be scrutinized within the wider context of sustainable malaria control and elimination strategies. In addition to an increasing use of more sensitive molecular diagnostic methods to achieve elimination and prevent resurgence, surveillance systems will hinge on which strategy is best suited to identify these asymptomatic infections.
Reactive case detection, whereby household members, neighbors, and other contacts of index cases are screened for infection and treated with antimalarial drugs [3], is a strategy recommended by the WHO. Despite the WHO's recommendation of RACD [5] and its use, knowledge gaps still exist in its implementation. These include questions to do with standard metrics on the coverage of screening needed to affect transmission, optimal target populations, timing, and frequency. Among the Asia Pacific Malaria Elimination Network (APMEN) partner countries, some countries have reported using a specific screening radius (maximum of 2.5 km) around index cases [6]. Procedures to determine neighboring households for investigation have included screening individuals residing within a particular radius of the index case and choosing a definite number of neighbors or households for follow-up [7]. Another study recommended the use of tablets loaded with satellite images from study sites to determine the proportion of households that should be tested [8]. At the individual country level, limited practical know-how exists to inform control programs. RACD can, however, be better executed when operational experiences from different countries are brought together to inform country-specific needs. Hence, the objective of this review is to discuss and highlight RACD as a recommended strategy for the detection and elimination of asymptomatic malaria as it pertains in different countries.
Few RACD reviews and meta-analyses that include research published up to the year 2022 currently exist. Newby et al. [9] and Perera et al. [10] reviewed the literature up to 2018 and 2019, respectively. Subsequently, a review by Stresman et al. [11] and a meta-analysis by Deen et al. [12] (confined to the Greater Mekong Sub-region) up to 2019 and 2021, respectively, were undertaken. A review and meta-analysis will expand the frontiers of current knowledge for programs geared towards malaria elimination by this strategy.

Data Sources and Search Strategy
A systematic review and meta-analysis of RACD strategies for malaria control and elimination was conducted per the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [13]. Search terms (combined free text and keywords) included "malaria and reactive case detection", "contact tracing", "focal screening", "case investigation" and "focal screen and treat". Relevant studies published between January 2010 and September 2022 were identified mainly through PubMed and Google Scholar. The retrieved studies were manually screened to identify the relevant studies.

Study Selection
Eligibility for inclusion was determined according to the following selection criteria: reported results of the RACD strategy (including number of cases followed, number of persons traced, number of new cases detected). The following information (Table 1) was extracted: (1) year of publication; (2) country of study; and (3) time period. Studies restricted to simulation, modeling, resampling algorithms, and articles that did not discuss RACD specifically were excluded. Review articles were also excluded from the analysis. Fifty-four (54) studies were systematically reviewed. Of these studies, 7 met the eligibility criteria based on risk of malaria infection in individuals living with an index case < 5 years old, 13 met the eligibility criteria based on risk of malaria infection in an index case household member compared with a neighbor of an index case, and 29 met the eligibility criteria based on risk of malaria infection in individuals living with index cases, and were included in the meta-analysis. A summary of the RACD strategies and outcomes, including the timing of the RACD, the baseline study participant characteristics, the RACD households, and the screening radius are shown in Table 1 and Supplementary Table S1.

Data Extraction and Assessment of Study Quality
Duplicates were removed with EndNote Reference Library. Studies that met the eligibility criteria were included after going through careful screening and evaluation by two reviewers (EKA and FTA). Studies were initially considered depending on their titles and abstracts, and this was followed by a thorough review of the complete article for relevance. To clarify any disagreements, a third investigator (SAS) was involved.

Statistical Analysis
Statistical analysis was carried out using MedCalc 20.0 (MedCalc Software Ltd., Ostend, Belgium). The findings from the pooled studies were reported as odds ratios with their confidence intervals from a pooled fixed-effect model. A Forest plot was generated from the outcomes to graphically represent the results. Publication bias and heterogeneity were tested using Egger's test and Higgins I 2 [67], respectively. For I 2 , values between 25% and 50% indicated low heterogeneity, values between 50% and 75% indicated moderate heterogeneity, and values greater than 75% indicated severe heterogeneity. A p-value of ≤0.05 was considered significant in all cases.

Search Result
In all, 115 studies were identified through PubMed and 284 studies were identified though Google Scholar (Figure 1). Following the removal of duplicates, 260 studies were screened, and this left 154 studies for full assessment. Of these, 100 studies were excluded because they concerned strategies other RACD (e.g., simulation, modeling). The remaining studies were carefully evaluated in detail, resulting in a total of 54 studies for qualitative synthesis and 7, 13, and 29 studies for quantitative synthesis.

Study Characteristics and Quality Assessment
All 7, 13, and 29 respective studies used for the meta-analysis involved real RACD strategies. In all, the studies included a total of 8965 index cases leading to 90,940 contacts (Supplementary Table S1). While 16 studies used only RDT to test contacts, 2 studies used RDT/PCR, 1 study used LAMP/PCR, 1 study used microscopy/PCR, 6 studies used PCR, 1 study used microscopy, 2 used studies LAMP, and 5 studies did not report the diagnostic method used (Supplementary Table S1). Index case demographics and individual study characteristics are shown in Supplementary Table S1. Although some of the included studies had a higher risk of bias than others, sensitivity analyses that excluded those studies showed significant difference. Quality assessment was independently carried out by two reviewers, and any disagreements were resolved by discussion.

The Risk of Malaria Infection in a Neighbor of an Index Case Compared with an Index Case Household Member
The pooled results from 13 studies indicated that neighbors of index cases were 0.352 [0.301-0.412] times more likely to have a malaria infection relative to index case household members, and this was statistically significant (p < 0.001) (Figure 3). The risk of infection from the pooled fixed effect model ranged between 0.029 and 4.062, and, from the results, high variation heterogeneity chi-square = 285.247, (p < 0.0001) I 2 = 95. 79  The pooled results from 13 studies indicated that neighbors of index cases were 0.352 [0.301-0.412] times more likely to have a malaria infection relative to index case household members, and this was statistically significant (p < 0.001) (Figure 3). The risk of infection from the pooled fixed effect model ranged between 0.029 and 4.062, and, from the results, high variation heterogeneity chi-square = 285.247, (p < 0.0001) I 2 = 95. 79 [94.81-96.96] (assessed with Egger's Test, intercept = 2.989, p = 0.903; test for overall effect, Z = 2.533, (p = 0.011)was observed.   [24,26,33,[36][37][38]41,43,49,53,54,58,64]. SE: Standard Error, Log(odds ratio) values with "-" are negative values.

Risk of Malaria Infection Living with an Index Case < 5 Years Old
The risk of malaria infection in an individual living with an index case < 5 years old according to the pooled fixed effect model ranged between 0.240 and 18.520. The average risk of infection was 3.882 (1.526-4.617) according to the pool of seven studies that were statistically significant (p = 0.003) (Figure 4). According to the pooled results, high variation heterogeneity chi-square = 25.47, (p < 0.0001) I 2 = 93.89 [91.36-96.96]. SE: Standard Error, Log(odds ratio) values with "-" are negative values Figure 3. The risk of malaria infection in a neighbor of an index case compared with an index case household member [24, 26, 33, 36-38, 41, 43, 49, 53, 54, 58, 64].

Risk of Malaria Infection Living with an Index Case < 5 Years Old
The risk of malaria infection in an individual living with an index case < 5 years old according to the pooled fixed effect model ranged between 0.240 and 18.520. The average risk of infection was 3.882 (1.526-4.617) according to the pool of seven studies that were statistically significant (p = 0.003) (Figure 4). According to the pooled results, high variation heterogeneity chi-square = 25.47, (p < 0.0001) I 2 = 93.89 [91.36-96.96].  [15,20,37,43,47,51,62]. SE: Standard Error, Log(odds ratio) values with "-" are negative values.

Publication Bias
Publication bias was not assessed for the outcome in Figure 4 involving the 7 studies for which a funnel plot was not retrieved [68], but the other 2 outcomes had their publication bias assessed [69].

Sensitivity Analysis
A sensitivity analysis was performed for the risk of malaria infection in a neighbor of an index case compared with index case household members to deal with heterogeneity. After excluding five studies (Conner et al. [26], Stuck et al. [33], Aidoo et al. [37], Hustedt et al. [53], and Fontoura et al. [54]), the analysis was performed again to generate a forest plot ( Figure 5). A change from 0.352 [0.301-0.412] to 0.459 [0.294-0.715] in the final odds ratio was observed, and heterogeneity was reduced from 95% to 46% when these studies were excluded.

Publication Bias
Publication bias was not assessed for the outcome in Figure 4 involving the 7 studies for which a funnel plot was not retrieved [68], but the other 2 outcomes had their publication bias assessed [69].

Sensitivity Analysis
A sensitivity analysis was performed for the risk of malaria infection in a neighbor of an index case compared with index case household members to deal with heterogeneity. After excluding five studies (Conner et al. [26], Stuck et al. [33], Aidoo et al. [37], Hustedt et al. [53], and Fontoura et al. [54]), the analysis was performed again to generate a forest plot ( Figure 5). A change from 0.352 [0.301-0.412] to 0.459 [0.294-0.715] in the final odds ratio was observed, and heterogeneity was reduced from 95% to 46% when these studies were excluded.

Discussion
Malaria elimination calls for strategies aimed at reducing parasite reservoirs, both in hypoendemic and hyperendemic areas. RACD strategies are utilized primarily by malaria elimination programs and have been implemented in Asia, South America, and  [24,36,38,41,43,49,58,64]. SE: Standard Error, Log(odds ratio) values with "-" are negative values.

Discussion
Malaria elimination calls for strategies aimed at reducing parasite reservoirs, both in hypoendemic and hyperendemic areas. RACD strategies are utilized primarily by malaria elimination programs and have been implemented in Asia, South America, and Africa (with mixed results). RACD application in locations in Africa or elsewhere may differ by way of coverage as a result of differences in human behavior, settlement patterns, and epidemiology of malaria transmission. RACD may provide a strategy to detect malaria hotspots. However, challenges may exist due to low parasite reservoir coverage despite large operational efforts.
The RACD studies under review examined factors relevant to malaria elimination, including diagnosis, infection reservoir (asymptomatic, submicroscopic), prevalence, surveillance, and elimination strategy. Data pooled from the different RACD studies were utilized to calculating the average risk of malaria infection among household members and neighbors. The pooling of quantitative data from regions with varying malaria endemicity may not truly reflect the risk of infection for a particular region as this will be affected by factors such as vector abundance, socioeconomic circumstances, malaria health policies in the regions, etc., and thus may differ accordingly.
In endemic regions, malaria transmission is often seasonal [70] and heterogeneous [71]. To determine whether the spatial targeting of high-risk areas is of concern, malaria control and elimination programs must understand the extent of the community-, district-, and regional-level spatial heterogeneity of malaria risk. By identifying hotspot locations, it is possible to quantify spatial risk and identify the particular risk factors that may be causing an increase in infection risk in those places. Malaria distribution in endemic countries is highly heterogeneous. From our meta-analysis, the distinct epidemiological features of malaria in different countries could be related to the significant heterogeneity in different transmission settings. Since RACD is best suited for hypoendemic areas, it could be possible that when a local area gets closer to eliminating local malaria, increased clustering is anticipated, and that these clusters may show particular features that are challenging to control. A RACD strategy that entails distinct transmission processes for local elimination can be utilized to further reduce malaria clusters in hypoendemic areas. However, in countries with a high malaria burden, routine strategies for case reduction can be continued or scaled up [72]. Per the results of the meta-analysis, the risk of malaria infection in individuals living with an index case, malaria infection risk in neighbors of an index case compared with index case household members, and malaria risk in individuals living with an index case < 5 years old were significant.
Using spatial extent or screening radius as a RACD implementation metric, numerous studies have indicated a variety of strategies [32,37,50,61]. While some investigations only involved household members [46], others expanded their investigation to include index case neighbors within varying distances [50,53]. Malaria transmission has been reduced as a result of interventions aimed at both entire communities and households with known malaria infections [30,73]. The majority of the positive cases identified by RACD were geographically found near index case households. Many African cities show a clear trend of increasing malaria transmission from urban to peri-urban to rural settings [74]. The spatial extent of RACD in urban Lusaka, where population density is higher and more people are tested, was associated with a higher probability of identifying additional cases within 250 m of uninhabited land [25]. Urban malaria transmission may vary depending on an assortment of factors, such as vector breeding sites, local vector species, location, human movement patterns, waste management, socioeconomic factors, land use, climate, and local malaria intervention programs [75]. In cases of low transmission, strategies that target the household are thought to be operationally sustainable. A RACD strategy targeting households offers a practical programming alternative given that achieving elimination would require sustained efforts over the long run.
Secondary infections are detected where neighbors have been involved in a RACD strategy, and some studies have shown significant declines in positivity as distances in-crease [8,32]. However, the operational practicality is constrained by the low numbers of secondary infections detected in field investigations, which necessitates screening a large number of people [76]. Here, we provided evidence to back up the notion of infection clustering in neighborhoods, which in turn makes the inclusion of neighboring households a part of the RACD strategy. Nevertheless, the ideal screening radius from the index case household to the neighboring households as part of RACD still remains unclear. The determination of a screening radius is informed by the local malaria epidemiology (including vector dispersal) and logistical capacity [5]. Targeting high household coverage over a smaller screening radius in unsprayed areas would enhance efficiency and responsiveness and help establish an ideal screening radius. While in Swaziland, a 1 km screening radius was operationally demanding, in some APMEN partner countries a maximum screening radius of 2.5 km was covered.
According to Sturrock et al. [8], RACD implementation metrics, such as response time, indicated that the odds of detecting secondary infections were much higher if RACD was triggered within a week of presentation of the index case. Such rapid response ensures that asymptomatic cases are identified prior to becoming symptomatic and seeking treatment. This allows only a limited opportunity to infect mosquitoes. While performing RACD, Aidoo et al. [37] screened 1280 individuals living in 413 households., and in the 100,000 individuals in the catchment area, an estimated total of 12,900 infections were present at any time. RACD identified a total of 144 additional cases in index case and neighbor households. However, the screening of 1000-2400 individuals would have led to the identification of 200-500 secondary cases each month. This translated into 1.5-3.9% of all asymptomatic infections relative to the estimated total of 12,900 infections in the catchment population of the hospital. To estimate the overall number of infections identified during an extended period, such as a whole year, it would be necessary to have a thorough understanding of the duration of infections, the total number of people infected over time, and the temporal variations in transmission foci. The odds of finding secondary cases were lower when the index household was sprayed. Where data on neighboring households are not available, spraying is normally directed at predefined locations rather than individual houses. Since index households could serve as indicators for spraying locations, its protective effect could possibly be extended to neighboring households, leading to a reduction in the local vector density. In addition, it is possible that infections in the sprayed regions are acquired while engaging in recreational or work-related activities outside the house. RACD should only be initiated in receptive areas where there is a likelihood of transmission around the vicinity of the index case. In these locations, RACD should be employed irrespective of whether a case has been imported or locally transmitted because imported cases can result in local transmission [3]. While some RACD studies concentrated only on symptomatic individuals, others focused on both symptomatic and asymptomatic individuals. In all these RACD studies, many more asymptomatic infections (mostly sub-microscopic) were identified compared with symptomatic infections. In hypoendemic areas, asymptomatic infections are common, the majority of which are sub-microscopic [77] and constitute up to 50% of human-to-mosquito transmissions [78]. Nevertheless, it is possible that hotspots of mainly asymptomatic infections were missed because symptomatic infections (index cases which triggered RACD) and asymptomatic infections do not necessarily overlap spatially [79]. Asymptomatic carriers can act as reservoirs of infection due to their failure to seek treatment [80], further derailing elimination efforts. The challenges of asymptomatic infections are also compounded by diagnostic limitations [81]. RACD strategies employing malaria diagnostic tools of high sensitivity supplement the everyday passive case detection in hypoendemic settings and may be used in malaria elimination programs. Thus, alternative diagnostic tools that are more field-friendly and sensitive, such as highly sensitive RDTs (HS-RDTs) or loop-mediated isothermal amplification (LAMP), need to be assessed and used for the identification of sub-microscopic/asymptomatic malaria. In lieu of this, presumptive treatment targeting vulnerable individuals during the high transmission sea-sons may be a strategy that can help address the challenge of low parasitaemia missed by RDTs and poor RACD screening coverage [82].
The implementation of RACD differed with the different subgroups tested (e.g., index case household only, index case household versus neighbors, index case household versus neighbors plus control group), the number of index cases that triggered RACD, and the radius covered. Though most studies reported adhering to case investigation data collection guidelines, there was a collection of variables that each study decided to analyze, and this reflected the tailoring of investigations to local conditions and capacies. A fully operational surveillance and response system is required for malaria elimination, in addition to at least three consecutive years with no indigenous malaria cases [83]. It has long been acknowledged that the elimination of malaria requires effective malaria surveillance [83]. With a workforce committed to surveillance at all levels of the health system [84] and integrated response mechanisms [85], countries that have successfully eliminated malaria have typically used a combination of effective passive case detection (PCD) [86] and active case detection (ACD) activities [84]. According to China's 1-3-7 framework, malaria cases must be reported within one day, cases must be investigated within three days, and foci investigation and increased preventive measures must be conducted by day seven [87]. Depending on the risk and endemicity levels, different responses are recommended, including "intensified surveillance and response" in border areas, "passive surveillance during the transmission season and active surveillance targeting transmission foci" in zones with seasonal malaria, and "active and passive surveillance, with special attention to mobile populations," in areas with higher incidence [88]. The Immediate Disease Notification System (IDNS), a surveillance system with notifiable disease system and surveillance outputs that swiftly transmit to a team to trigger a response, is part of the malaria surveillance system in Swaziland [89]. It would be ideal for countries that have eliminated malaria to share their individual experiences. In endemic countries, sharing data on population movements across borders will make it easier to eliminate the disease. This strategy, in our opinion, would be beneficial for regional cooperation in the fight against malaria [89].
Operational gaps exist concerning which RACD approach is most effective at reducing and preventing transmission. This review recommends the use of standardized data metrics in assessing the RACD screening radii, the number of individuals tested, the completeness of geographical coverage during RACD, and the timeliness of the incorporated activities. The integration of more reliable data and the development of maps of foci of transmission by geo-locating cases to the households can support the targeting of these vulnerable individuals and their neighborhoods with malaria interventions [90,91].

Conclusions
Despite the WHO surveillance benchmarks for elimination in endemic regions and the operational know-how of affected countries, it is not likely that recommendations for one setting will apply to all others across diverse epidemiological areas. The choice of whether or not to use a RACD strategy is dependent on an assortment of factors including local transmission intensity, cost, operational feasibility, and population receptiveness. As countries progress towards elimination, malaria programs should prioritize case investigation and undertake RACD to identify remaining reservoirs of infection.