Dihydroartemisinin-Piperaquine Chemoprevention and Malaria Incidence After Severe Flooding: Evaluation of a Pragmatic Intervention in Rural Uganda

Abstract

Background

Malaria epidemics are a well-described phenomenon after extreme precipitation and flooding. Yet, few studies have examined mitigation measures to prevent post-flood malaria epidemics.

Methods

We evaluated a malaria chemoprevention program implemented in response to severe flooding in western Uganda. Children aged ≤12 years from 1 village were eligible to receive 3 monthly rounds of dihydroartemisinin-piperaquine (DP). Two neighboring villages served as controls. Malaria cases were defined as individuals with a positive rapid diagnostic test result as recorded in health center registers. We performed a difference-in-differences analysis to estimate changes in the incidence and test positivity of malaria between intervention and control villages.

Results

A total of 554 children received at least 1 round of chemoprevention, with 75% participating in at least 2 rounds. Compared with control villages, we estimated a 53.4% reduction (adjusted rate ratio [aRR], 0.47; 95% confidence interval [CI]: .34–.62; P < .01) in malaria incidence and a 30% decrease in the test positivity rate (aRR, 0.70; 95% CI: .50–.97; P = .03) in the intervention village in the 6 months post-intervention. The impact was greatest among children who received the intervention, but decreased incidence was also observed in older children and adults (aRR, 0.57; 95% CI: .38–.84; P < .01).

Conclusions

Three rounds of chemoprevention with DP delivered under pragmatic conditions reduced the incidence of malaria after severe flooding in western Uganda. These findings provide a proof-of-concept for the use of malaria chemoprevention to reduce excess disease burden associated with severe flooding.

The impact of climate change on the incidence of vector-borne diseases, including malaria, is an issue of substantial public health importance [1, 2]. In addition to rising surface temperatures, the manifold effects of global climate change also entail an increased frequency of weather extremes such as droughts and floods [3, 4]. Nearly half of global disasters over the past 2 decades were attributable to extreme precipitation and flooding, including 64% of events in the Africa region [5]. Populations in developing countries are particularly at risk of adverse health consequences from floods [6]. For many of the same reasons, these are also areas where malaria is endemic.

Unlike the immediate impacts of flooding, malaria epidemics emerge after the acute phase of the crisis has passed [7]. Heavy precipitation is thought to flush established larval habitats—an effect that can result in short-term reductions in disease transmission [8]. As floodwaters recede, however, malaria vectors rapidly reestablish breeding sites, and a surge in disease may occur months after the disaster [9–11]. In a previous study, we observed a 30% increase in malaria test positivity and a 40% increase in malaria-related hospitalization after severe flooding in western Uganda [12]. While the peak risk was seen approximately 3 months after the flood, risk remained elevated for nearly a year. Even if temporary, the burden of increased malaria transmission is substantial, particularly among young children [13, 14]. Households also incur economic and opportunity costs due to care-seeking [15, 16]. Despite these impacts, few studies have examined post-flood malaria mitigation measures.

Chemoprevention involves the administration of medication to prevent disease or infection. Malaria chemoprevention is increasingly used for disease control in malaria-endemic regions and can be delivered in various ways, including mass drug administration (MDA) and intermittent preventive treatment of specific high-risk populations. In 2012, the World Health Organization (WHO) recommended the implementation of seasonal malaria chemoprevention (SMC), essentially short periods of MDA, in areas where malaria transmission is highly seasonal [17]. SMC has been shown to be highly effective at both the individual and population levels [18, 19] and is now implemented in 13 countries, with more than 21 million children receiving at least 1 dose in 2019 [20].

Similar approaches applied to settings where transmission is perennial have demonstrated effectiveness [21]. More widespread adaptation of routine chemoprevention programs in young children is hampered by a rapid return to baseline levels of transmission as well as concerns about the emergence of artemisinin resistance and questions of sustainability. At present, it is unclear in what context various forms of chemoprophylaxis may be of most benefit in such settings [22]. Yet, recent studies have demonstrated that more targeted uses of chemoprevention may offer opportunities to reduce excess disease transmission and limit morbidity and mortality in the short term [23, 24].

In May 2020, a severe flooding event occurred in Kasese District, a rural, malaria-endemic area of western Uganda. The event displaced more than 100 000 residents from their homes, particularly those along the Mubuku and Nyamwamba rivers (Figure 1) [25]. We hypothesized that MDA with dihydroartemisinin-piperaquine (DP), which has a relatively simple dosing schedule, long post-treatment prophylaxis period, and established safety profile, might serve as a valuable tool to mitigate a post-flood surge in malaria [26, 27]. To test this hypothesis, we carried out a pragmatic intervention aimed at reducing excess malaria incidence associated with flooding. Here, we assess the impact of this program, comparing the intervention village with 2 neighboring villages where no intervention was deployed.

METHODS

Study Site

Izinga village is located in the Maliba subcounty of Kasese District (0.3006 N, 30.1059 E). Geographically, Izinga occupies an area of approximately 1.3 km² and is situated between the Mubuku River to the west and Kitajuka River to the east, with the 2 rivers intersecting in both north and south (Figure 2). Compared with the surrounding terrain, the area is relatively flat and low-lying. According to the 2018 census conducted by community health workers (CHW), the village population is comprised of 188 households with 1118 residents of whom 452 (38.1%) were reported to be children aged <12 years. Population projections available at the time of the flooding assumed a flat growth rate of 3.0% per year, which was likely an underestimate [28].

The climate in the Kasese District permits year-round malaria transmission with seasonal peaks in January and July following the end of the wet seasons. Plasmodium falciparum accounts for the overwhelming majority (>95%) of infections [29, 30]. The most recent malaria indicator surveys undertaken in the Mid-Western region (2014–2015) and Tooro subnational region (2018–2019), which include the study area, reported P. falciparum parasitemia rates (PfPR) of 17.4% and 7.3%, respectively [31, 32]. As part of an ongoing study of malaria, however, we had conducted a household survey of children aged 2–10 years in Izinga approximately 2 months prior to the flooding. In our survey, using rapid diagnostic tests (RDTs), we identified P. falciparum parasitemia in 18 of 60 children, which resulted in a weighted estimate of PfPR_2-10 of 30.0% (95% confidence interval [CI]: 24.8%–35.8%).

Intervention

In response to the severe flooding that occurred on 7 May 2020, members of the Mbarara University of Science and Technology–University of North Carolina at Chapel Hill Research Collaboration coordinated with a local nongovernmental organization (NGO) to plan and carry out a chemoprevention intervention in Izinga where the collaboration has a long-standing partnership. Based on previous work that showed a transmission peak approximately 3 month post-flooding, a written proposal to provide 3 rounds of monthly DP as chemoprevention to children aged ≤12 years was submitted to and approved by the Kasese District Health Officer [12]. The NGO subsequently led community sensitization meetings to inform residents of the proposed activities in an effort to optimize engagement and participation.

The first round of chemoprevention was launched on 10 June 2020, approximately 1 month after the flood. CHWs led field staff to each household in their respective area of responsibility. When an eligible household (eg, at least 1 child aged ≤12 years in the home) was reached, field staff described the program objectives and activities to the adult caregiver who was asked to provide verbal consent to participation. Staff administered a brief questionnaire that elicited responses about displacement due to the flooding, incident malaria cases since the flood, and bed net ownership and use in order to identify additional needs and guide future response. Eligible children were weighed to determine dosing using established guidelines [33]. Children who weighed less than 5 kg were excluded from the intervention. Staff administered the first dose of DP (Duo-Cotecxin, Holly-Cotec Pharmaceuticals, Beijing, China), and children were observed for 15 minutes to monitor for adverse reactions including vomiting. If the child vomited, staff provided the caregiver with a replacement dose and instructions to readminister the following day. Caregivers were provided with adequate tablets of DP and instructions to administer doses on days 2 and 3. Similar programs were carried out starting 16 July 2020 and 5 September 2020 (Figure 3). CHWs monitored participants for adverse events following each round. Total costs of the program, including field staff, supplies, and medication were summarized.

Program Evaluation

We selected 2 neighboring villages, Kakindo and Kanyamingo, as comparator villages to assess the effectiveness of the intervention in Izinga. These villages were selected due to their close proximity (Figure 2) and broadly similar demographic and geographic characteristics (Supplementary Table 1, Supplementary Figure 1). While these villages were also impacted by the flooding, no malaria-specific response was carried out until a government-led long-lasting insecticidal nets (LLIN) distribution campaign took place in mid-August 2020. To assess for changes in malaria incidence, we abstracted clinical information from registers at the 5 nearest health facilities and local CHWs conducting integrated community case management of malaria, which would be expected to capture the vast majority of malaria care-seeking. For each patient who presented from 1 of 3 villages, we recorded the date of the visit, the age and sex of the patient, and the malaria RDT result. Data were collected from November 2019, a time period 6 months prior to the flood, through November 2020, approximately 6 months after the flood. All information was abstracted in and uploaded to a secure electronic database (eg, REDCap) using smart phones with cellular Internet connectivity [34].

Statistical Analyses

The primary outcome of interest was the incidence of clinical malaria, defined as the number of individuals who presented to one of the catchment health facilities with a positive RDT result per village per week, using CHW census information to provide village-level population offsets. Our primary explanatory variable was calendar time, which we divided into “pre-intervention” and “post-intervention” periods. We first graphically depicted trends in the crude number of febrile illness visits (eg, RDT negative) and malaria cases (eg, RDT positive) over calendar time to examine patterns of care-seeking in relation to the flooding. We then compared malaria incidence before and after the flood by fitting generalized additive models with a negative-binomial response using the mgcv package in R [35]. Incidence was modeled for each village and age category (eg, <5 years, 5–12 years, ≥12 years). Effect size was estimated using a difference-in-differences approach, averaging across the effect size of age groups within a village and the control villages. As a secondary outcome, we examined the test-positivity rate (TPR), defined as the number of positive RDT results per 100 tests performed using similar models. The number of cases averted was calculated as the difference between the incidence in Izinga following the beginning of the intervention and the incidence without the intervention as derived from the 2 control villages (Supplementary Table 2, Supplementary Figure 2). Estimates were made using the predict.gam() function in the mgcv package.

Ethical Approvals

The University of North Carolina at Chapel Hill, Mbarara University of Science and Technology, and Uganda National Council for Science and Technology institutional review boards provided ethical approval.

RESULTS

Over the course of the intervention, 554 children at 157 unique households received at least 1 course of chemoprevention; a number that is approximately 20% higher than reported on the most recent CHW census. When a parent or guardian was able to be located, participation was high. Assuming that the actual eligible population was 554 children, rather than our initial estimate of 452 children, we achieved coverage rates of 79.8%, 76.9%, and 77.6% in each round, respectively (Figure 3). No refusals were documented. More than 60% of participating children (n = 335) completed all 3 rounds, and nearly 75% of children (n = 413) completed at least 2 rounds (Table 1).

The median age of participating children was 7 years (interquartile range [IQR], 3–10). Baseline LLIN ownership and reported use as assessed during the first visit was well below WHO targets, which may be partly due to the flood event occurring toward the end of a 3-year distribution cycle [36]. More than half of participating households reported temporary displacement due to flooding, but generally for less than 1 week (3 days; IQR, 2–5). Households that participated in at least 2 rounds of the intervention reported longer periods of displacement (P < .01) and more frequent malaria episodes (P < .01) compared with those who only participated in a single round. Of the 1296 initial doses that were directly observed, the vast majority (1227, 94.7%) were tolerated without immediate vomiting. No serious adverse events, including allergic reactions, were reported over the intervention period.

In the 6-month period prior to flooding, the crude incidence of malaria was higher in Izinga when compared with Kanyamingo and Kakindo (Table 2) but was statistically similar in the adjusted model (adjusted rate ratio [aRR], 1.17; 95% CI: .84–1.48; P = .37 and aRR, 1.25; 95% CI: 1.00–1.57; P = .13, respectively). After the flooding, we observed increases in the weekly incidence and TPR in all villages but substantially larger increases in the 2 control villages compared with the intervention village (Figure 4). The incidence peaked approximately 2–3 months after the flood and remained elevated above the study period’s mean through the subsequent wet season in October. Using a difference-in-differences approach, the incidence of malaria following the intervention was significantly lower in the intervention village compared with the control villages (aRR, 0.47; 95% CI: .34–.62; P < .01), representing an approximate 53% reduction. As expected, the greatest reductions in malaria incidence were observed in the target group, children aged ≤12 years (aRR, 0.42; 95% CI: .29–.60; P < .01), with a smaller effect observed among older children and adults (aRR, 0.57; 95% CI: .38–.84; P < .01).

Similar findings were observed with the TPR. Prior to flooding, the TPR was similar between the intervention and control villages (Table 3). In the post-flood period, however, the crude TPR was significantly lower in Izinga compared with Kanyamingo and Kakindo (Supplementary Figures 3–5). In the difference-in-differences model, the adjusted risk ratio of a positive test was approximately 30% lower in the intervention compared with the control villages in the post-flood period (aRR, .70; 95% CI: .50–.97; P = 0.03).

We estimate that the chemoprevention with DP averted 318 cases (95% CI: 294–342) between 1 June 2020 and 30 November 2020 (Supplementary Figures 5 and 6). While children aged 12 years represent the largest share of averted cases (217, 68.2%), the model also predicted a substantial decrease in cases among individuals aged >12 years who were not eligible for the intervention (Figure 5). Total program costs were $5046 (Supplementary Table 3) or $3.89 per course of DP delivered. Using the above estimate, these costs are $15.87 (95% CI: $14.75–$17.16) per case averted.

DISCUSSION

Three rounds of chemoprevention with DP targeted to children aged ≤12 years resulted in substantial reductions in the incidence of malaria after severe flooding in western Uganda. The high rate of participation and low rate of adverse events suggest that the intervention was both feasible and broadly acceptable. Furthermore, the resources required to deliver the intervention were time-limited and relatively modest. Together, these findings, notable for the fact that they were obtained from a pragmatic intervention conducted in the context of a natural disaster amidst an ongoing global pandemic, provide a proof-of-concept for the use of malaria chemoprevention to reduce excess disease burden associated with severe flooding.

Our estimation of a 53% reduction in disease incidence in the 6 months of observation after flooding is somewhat less than that observed in clinical trials [21]. These differences may be attributable to difference in context and design but may also have resulted from lower population coverage rates. While we were able to reach approximately 75% of eligible children each round, only 60% completed all 3 rounds. This suggests that a substantial proportion of children were at risk of infection for at least 30 days during the periods of peak transmission intensity. Modeling studies have shown that coverage is a critical component in determining the effect of MDA on percentage reduction in parasite prevalence [37]. The reduced effect may also be due to imperfect adherence to the full, 3-day course of DP. Given logistical constraints, we were only able to observe the first dose and relied on caregivers to administer subsequent doses. Despite these issues, the observed effects are comparable with the findings from “real-world” programs conducted in areas of seasonal transmission [18]. Similarly, our estimates of cost-effectiveness, both per course delivered and case averted, fall well within the range of previous estimates derived from SMC programs [38].

Our finding of sustained reductions in disease incidence and test positivity in the 3 months after the end of intervention is consistent with results from a study in Gambia that used a preseasonal MDA approach [39]. These results suggest that the intervention was effective in reducing the parasite reservoir in the eligible population, which had a prolonged impact on local transmission even well after the last round of chemoprevention [40]. We also observed smaller but significant reductions in malaria incidence among older children and adults who were not eligible to receive chemoprevention. The indirect effect on the larger population is likely due to the fact that school-aged children often represent a disproportionate burden of parasitemia and treatment resulted in a smaller parasite reservoir. Similar effects on the larger population have been observed in low-transmission settings, whereas more modest impacts were seen in high-transmission settings when chemoprevention programs were implemented among school-aged children [41–43].

This study has a number of unique characteristics and strengths, the most remarkable of which is the context under which the study was conducted. While the circumstances imposed by the flooding and coronavirus disease 2019 pandemic compressed our timelines, restricted access, and limited our ability to mobilize resources, they also allowed us to evaluate the program under the operational stresses expected to be encountered in disaster relief. The program also benefited from established partnerships with the affected communities that have been shown to be crucial factors associated with successful interventions in other settings [44].

Our study also has important limitations, foremost of which is the quasiexperimental design and reliance on routine health facility data. While we attempted to select control villages with similar characteristics that were also impacted by the flooding, it is possible that there were differences between the populations that may have differentially affected malaria risk in pre-flood periods. The use of the difference-in-differences approach, however, did allow us to control for some of the variability in the analysis. To minimize the risk of bias in the outcome measures, we assessed our outcome similarly in both the intervention and control areas by reviewing cases of incident malaria cases from nearby health facilities and CHW records. However, we may still have missed individuals who sought care at unregistered private sector drug shops, traditional healers, or referral centers outside the catchment area. This is unlikely to have confounded our analysis unless there was a difference in care-seeking preferences between villages, but we cannot rule this out. Second, the relatively small geographic area and close proximity of the villages may have permitted spillover of the intervention effect into neighboring areas. We note, for example, that 554 children received at least 1 course of chemoprevention, a number that is approximately 20% higher than the eligible population of children from the most recent census. The discrepancy is likely due, in part, to population growth that was more rapid than estimated but may also reflect children and caregivers from other villages, either by coincidence or purposefully, who were present on the days the intervention was delivered. Regardless, spillover of the intervention would be expected to bias the result toward the null, and we are reassured by the robustness of our findings. Last, the relatively short duration of follow-up limits our ability to assess long-term changes in disease incidence, including potential rebound effects.

CONCLUSIONS

Three monthly rounds of chemoprevention with DP resulted in significant reductions in P. falciparum incidence after severe flooding in western Uganda. While these results are limited by the small scale of the program, the intervention appears to be feasible, acceptable, and effective even under challenging conditions. The application of focused chemoprevention strategies, typically limited to highly seasonal transmission settings, merits further investigation as a means of reducing excess morbidity and mortality associated with natural disasters and other time-limited crises in areas of perennial transmission. Additional studies are needed to confirm our findings and refine criteria for use.

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.

Notes

Author contributions. Study conception and design: R. M. B., E. X., R. R., E. B., R. M., E. M. M. Funding: R. M. B. Study implementation: R. M. B., E. X., E. B., R. M., M. N., E. M. M. Data analysis: E. X., B. D. H., V. G., R. M. B. First draft of manuscript: E. X., B. D. H., V. G., A. B. M., R. M. B. Revisions: All.

Acknowledgments. We thank the residents of Izinga village who participated in the study. In addition, we recognize the support of the Kasese District Health Office, which facilitated study implementation. Deidentified individual data that support the results will be shared beginning 9 to 36 months following publication provided the investigator who proposes to use the data has approval from an institutional review board, independent ethics committee, or research ethics board, as applicable, and executes a data use/sharing agreement with the University of North Carolina at Chapel Hill.

Financial support. R. M. B. is supported by the National Institutes of Health (K23AI141764) and received additional funds from a Caregivers at Carolina Award made by the Doris Duke Charitable Foundation (award 2015213). R. M. B. and R. R. report a supply donation (via a nongovernmental organization [NGO]) from Medilink Uganda and donations to Relief Efforts (via an NGO) from the UNC Health Foundation. V. G. acknowledges support from the Eunice Kennedy Shriver National Institute of Child Health and Human Development under award P2C HD050924. E. X. acknowledges the support of a Benjamin H. Kean Travel Fellowship from the American Society of Tropical Medicine and Hygiene.

Potential conflicts of interests. R. M. B. reports a donation of malaria rapid diagnostic tests for studies from SD Bioline. R. R. reports serving on the advisory board of the Ugandan NGO for People’s Health and Economic Development Organization. M. N. reports serving as executive director of an NGO involved in the study. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References

Author notes