Negligible Impact of Mass Screening and Treatment on Mesoendemic Malaria Transmission at West Timor in Eastern Indonesia: A Cluster-Randomized Trial

Abstract Background Mass screening and treatment (MST) aims to reduce malaria risk in communities by identifying and treating infected persons without regard to illness. Methods A cluster-randomized trial evaluated malaria incidence with and without MST. Clusters were randomized to 3, 2, or no MST interventions: MST3, 6 clusters (156 households/670 individuals); MST2, 5 clusters (89 households/423 individuals); and MST0, 5 clusters (174 households/777 individuals). All clusters completed the study with 14 residents withdrawing. In a cohort of 324 schoolchildren (MST3, n = 124; MST2, n = 57; MST0, n = 143) negative by microscopy at enrollment, we evaluated the incidence density of malaria during 3 months of MST and 3 months following. The MST intervention involved community-wide expert malaria microscopic screening and standard therapy with dihydroartemisinin-piperaquine and primaquine for glucose-6 phosphate dehydrogenase–normal subjects. All blood examinations included polymerase chain reaction assays, which did not guide on-site treatment. Results The risk ratios for incidence density of microscopically patent malaria in MST3 or MST2 relative to that in MST0 clusters were 1.00 (95% confidence interval [CI], .53–1.91) and 1.22 (95% CI, .42–3.55), respectively. Similar results were obtained with molecular analysis and species-specific (P. falciparum and P. vivax) infections. Microscopically subpatent, untreated infections accounted for 72% of those infected. Conclusions Two or 3 rounds of MST within 3 months did not impact the force of anopheline mosquito-borne infection in these communities. The high rate of untreated microscopically subpatent infections likely explains the observed poor impact. Clinical Trials Registration NCT01878357.

transmission prevail with Plasmodium vivax [15]. This study offers a first evaluation of MST in a low-transmission area with P. falciparum and P. vivax. Antimalarial therapy limited to blood schizontocides includes artemisinin derivatives and partner drugs, which do not impact the latent reservoir of P. vivax hepatic hypnozoites [16,17]. In this cluster-randomized study, MST consisted of expert microscopic mass blood screening guiding the immediate administration of blood schizontocidal therapy together with hypnozoitocidal primaquine for patients diagnosed with P. vivax, and gametocytocidal therapy for those having active P. falciparum malaria. Control clusters received no MST interventions. We aimed to maximize MST impacts in assessing broader relevance where low-level transmission of both dominant Plasmodium species occurs.

Study Design and Location
Malaria in communities tends toward uneven distributions due to environmental, demographic, and socioeconomic factors [18][19][20]. This study, conducted during 2013, utilized an open-label, community-wide cluster-randomized controlled trial in Wewiku subdistrict, West Timor, Indonesia ( Figure 1). This subdistrict comprised 12 villages having 17 423 residents living in traditional coconut palm homes without electrical supply. A tropical climate occurs with a brief wet season (December-March) and extended dry season (April-November). The study coincided with peak malaria transmission during August to September. Annual parasite incidence was 72 and 124 per 1000 person-years during 2011 and 2012, respectively (Belu District Health Office, personal communication). Temperature ranges between 27°C and 35°C with average annual rainfall typically >700 mm. This lowland area (<150 m) includes coastal, savannah, paddy, and forest plantation ecosystems. Ditches, small streams, and semipermanent ground pools are primary anopheline larval habitats. Anopheles barbirostris dominates among suspected vector species, followed by Anopheles subpictus and Anopheles vagus. Limited government vector control during this study included the distribution of 800 insecticide-treated nets during 2011 (Belu District Health Office, personal communication). Because malaria prevalence was highest among coastal villages (Belu District Health Office, personal communication), 5 were selected for screening for inclusion in this study: Alkani, Lamea, Weoe, Seserai, and Weseben ( Figure 2).

Procedures
Randomization, intervention allocations and study procedures were explained to community leaders and household heads during a study socialization event. Household heads (n = 459) signed informed consent on behalf of their respective household members. MST3 intervention occurred monthly from June to August 2013, at 5-week intervals. MST2 occurred over the same period with a 10-week interval. MST0 residents were not screened. Schoolchildren cohorts were screened monthly from June to November 2013.
Treatment doses and adverse event (AE) monitoring was directly observed by staff at community health centers, health workers at integrated services posts, or elementary school teachers. Drug adherence was defined as taken completely as prescribed with witnessing, and occurred with >90% of cases. Local health centers provided treatment for alternative drugs (eg, quinine for pregnant women) or antipyretics, antiemetics, and antibiotics for AEs and other conditions. All treatments and AEs were reported and recorded daily to the research team.

Blood Examinations
Finger-prick blood films on glass slides were collected at arranged times and places. Approximately 250 µL of blood was collected in ethylenediaminetetraacetic acid tubes for downstream analyses and stored at −20°C. Thick and thin blood smears were stained 40 minutes with 3% Giemsa and examined using standard oil immersion light microscopy. Parasites were counted against 200 leukocytes and expressed per microliter assuming a leukocyte count of 8000/µL. One hundred ocular fields were examined on site before declaring a smear negative. Agreement between the field and laboratory reader was 0.75. These readings were blinded. DNA was extracted from whole blood using High Pure PCR Template Preparation kit (Roche Diagnostic) as per the manufacturer's instruction. The real-time PCR multiplex assay employing SYBR green used a Light Cycler Nano instrument to amplify the 18S ribosomal RNA gene [23]. Details on laboratory procedures are shown in the Supplementary Materials.

Statistical Analysis
The recruitment target was 1029 subjects per arm, considering a cluster design effect of 1.5, 5% significance, 80% power, and a 1:1 sample size ratio between intervention and control arms. A target of 115 children per arm would yield a power of 82% in detecting an estimated 50% reduction in malaria incidence following MST.
All analyses were performed using IBM SPSS Statistics for Windows version 23 (IBM, Armonk, New York). Change in prevalence was the primary endpoint, examined by comparing prevalence between intervention and control arms using generalized estimating equation modeling for cluster level and the χ 2 test for individual level [24].
Another primary endpoint was the risk ratio (RR) of P. falciparum or P. vivax incidence density between the MST3, MST2, and MST0 arms. Incidence was estimated by first microscopic infection during 6 months of observation of schoolchildren. Schoolchildren having microscopically patent parasitemia at enrollment were not analyzed, leaving 124, 57, and 143 malaria-negative children (MST3, MST2, and MST0, respectively; Figure 2). No significant differences were seen in demographic characteristics between 38 malaria-positive (excluded) and 324 malaria-negative (included) children. To adjust for possible clustering effect, the geometric mean of the cluster incidence RR was used to estimate intervention effect [25]. Cox proportional hazards regression calculated the hazard ratio of the individual-level incidence density between the intervention and control arms [26].
Ethical approval was obtained from the Health Research Ethics Committee, Faculty of Medicine, University of Indonesia Cipto Mangunkusumo Hospital (number 39/ H2.FI/ ETHICS/2013) in Jakarta, Indonesia. As the clinical trial authority, inspection by Indonesian Food and Drug Administration was conducted (October 2013).

Study Participants
Similar demographic characteristics occurred among both the cluster and individual levels among MST3, MST2, and MST0 residents ( Table 1). Fourteen community residents withdrew (Figure 2), 5 from MST3 and 9 from MST2, due to migration or illness.

Microscopic Examinations
Based on cluster analysis, malaria incidence densities (first event per person-year) were 1. 16 Table 2). Similar results were obtained with individual analyses (Figure 4A-C).

PCR Examinations
Cluster-based incidence density of malaria infections measured by PCR was 3.03, 3.83, and 2.67 infections per person-year for MST3, MST2, and MST0, respectively (  The table shows no significant difference of cluster-based malaria incidence density of both species among all arms. Risk ratio was calculated from the ratio of incidence density in the intervention and control arms. Geometric mean of cluster incidence was used to adjust the possible clustering effect.

DISCUSSION
This study demonstrated that MST may have little or no impact on malaria transmission in endemic communities where the majority of malaria infections cannot be detected by standard point-of-care diagnostics. Unrestricted malaria infection occurred with 2 or 3 rounds of MST, despite screening coverage >80%-higher than that typically attained in practice. Dominant subpatent and asymptomatic reservoirs, along with the latent hypnozoite reservoir of P. vivax, appear to have sustained transmission during the 3 months of intervention and the following 3 months. Thus, despite far lower transmission intensity, expert microscopy, and the use of primaquine as a transmission-blocking gametocytocide and hypnozoitocide, this study demonstrated similar results in the African setting of intense P. falciparum transmission diagnosed by RDT and treated with artemisinin-based combination therapy alone [5][6][7][8][9][10][11]. Our attempt to optimize microscopy-based MST for an Asian setting thus appears to be proven futile. A 2-fold statistically significant decrease between the MST3 and MST2 arms in P. falciparum prevalence was observed at the last round of MST by both microscopy and PCR diagnosis. This may be attributed to the greater efficiency of diagnosis of P. falciparum relative to P. vivax (ie, 50% vs 20%; Table 3). Furthermore, the decrease of gametocyte carriers of P. falciparum in MST3 relative to MST2 ( Figure 3C) did not translate to diminished risk of new infection. In P. vivax-infected subjects, 3 recurrences appeared within 2 months among 80 residents given antirelapse primaquine therapy (3.6%). In the cohort of schoolchildren, 10.5% (9/86) had recurrences within 5 months. All subjects received 0.25 mg/kg per national treatment guidelines; this relatively low dose may be inadequate [27]. The significant effort and expense made to safely diminish hypnozoite-borne blood infection in these communities with primaquine therapy exerted no discernible impact on risk of acute patent P. vivax, likely due to the inability to diagnose and treat the majority of subpatent infections. The latent reservoir of vivax malaria was not significantly impacted by MST that included a hypnozoitocide.
There are several important limitations to this study. Presumptive radical cure of the schoolchildren cohorts-to eliminate subpatent malaria infections-did not occur. These preinfections, rather than biting infectious mosquitoes, may account for some of the new infections detected and counted. Nonetheless, these were assumed to be balanced among the randomized arms, and therefore minimally impacting the MST effectiveness endpoints of RR. Another weakness may be the relatively small number of clusters per MST intervention (5 or 6), thus limiting the statistical power (<80%) to detect relatively small differences. Subtle but real differences are unimportant in the context of practice, where impacts must be relatively large for justifying costly and labor-intensive MST operations.
Additionally, the enrollment targets based on detecting a 50% reduction were not met. The RR near unity pointed to little difference of impact between arms, and the proportionate CI supported this finding (Table 2). Thus, this negative result was not likely caused by the small sample size [28,29]. The shortfall in sample size did not statistically impede the assessments.
In summary, 2 or 3 rounds of MST had no impact on the force of malaria infections in these communities. Effective MST requires improved sensitivity in point-of-care diagnostics [30]. Both highly sensitive antigen-detecting RDTs and near-patient malaria LAMP are now commercially available [14,31]. These tools may measurably improve the efficacy of MST [14,30]. Additionally, combining MST with other interventions that minimizing human-mosquito contact, may demonstrate detectable impacts. Finally, MST undertaken here exploited the availability of point-of-care diagnostics for G6PD deficiency and included hypnozoitocidal therapy for an underdiagnosed, understudied, and undertreated clinical and public health problem [32][33][34][35]. While we demonstrated the operational feasibility of safely including hypnozoitocidal primaquine therapy, inadequate infection diagnostics diminished possible impacts.

Supplementary Data
Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.