Design of effective outpatient sentinel surveillance for COVID-19 decision-making: a modeling study.

BACKGROUND: Decision-makers impose COVID-19 mitigations based on public health indicators such as reported cases, which are sensitive to fluctuations in supply and demand for diagnostic testing, and hospital admissions, which lag infections by up to two weeks. Imposing mitigations too early has unnecessary economic costs while imposing too late leads to uncontrolled epidemics with unnecessary cases and deaths. Sentinel surveillance of recently-symptomatic individuals in outpatient testing sites may overcome biases and lags in conventional indicators, but the minimal outpatient sentinel surveillance system needed for reliable trend estimation remains unknown. METHODS: We used a stochastic, compartmental transmission model to evaluate the performance of various surveillance indicators at reliably triggering an alarm in response to, but not before, a step increase in transmission of SARS-CoV-2. The surveillance indicators included hospital admissions, hospital occupancy, and sentinel cases with varying levels of sampling effort capturing 5, 10, 20, 50, or 100% of incident mild cases. We tested 3 levels of transmission increase, 3 population sizes, and conditions of either simultaneous transmission increase or lagged increase in the older population. We compared the indicators' performance at triggering alarm soon after, but not prior, to the transmission increase. RESULTS: Compared to surveillance based on hospital admissions, outpatient sentinel surveillance that captured at least 20% of incident mild cases could trigger an alarm 2 to 5 days earlier for a mild increase in transmission and 6 days earlier for a moderate or strong increase. Sentinel surveillance triggered fewer false alarms and averted more deaths per day spent in mitigation. When transmission increase in older populations lagged the increase in younger populations by 14 days, sentinel surveillance extended its lead time over hospital admissions by an additional 2 days. CONCLUSIONS: Sentinel surveillance of mild symptomatic cases can provide more timely and reliable information on changes in transmission to inform decision-makers in an epidemic like COVID-19.

Perennial malaria chemoprevention with and without malaria vaccination to reduce malaria burden in young children: a modelling analysis.

BACKGROUND: A recent WHO recommendation for perennial malaria chemoprevention (PMC) encourages countries to adapt dose timing and number to local conditions. However, knowledge gaps on the epidemiological impact of PMC and possible combination with the malaria vaccine RTS,S hinder informed policy decisions in countries where malaria burden in young children remains high. METHODS: The EMOD malaria model was used to predict the impact of PMC with and without RTS,S on clinical and severe malaria cases in children under the age of two years (U2). PMC and RTS,S effect sizes were fit to trial data. PMC was simulated with three to seven doses (PMC-3-7) before the age of eighteen months and RTS,S with three doses, shown to be effective at nine months. Simulations were run for transmission intensities of one to 128 infectious bites per person per year, corresponding to incidences of < 1 to 5500 cases per 1000 population U2. Intervention coverage was either set to 80% or based on 2018 household survey data for Southern Nigeria as a sample use case. The protective efficacy (PE) for clinical and severe cases in children U2 was calculated in comparison to no PMC and no RTS,S. RESULTS: The projected impact of PMC or RTS,S was greater at moderate to high transmission than at low or very high transmission. Across the simulated transmission levels, PE estimates of PMC-3 at 80% coverage ranged from 5.7 to 8.8% for clinical, and from 6.1 to 13.6% for severe malaria (PE of RTS,S 10-32% and 24.6-27.5% for clinical and severe malaria, respectively. In children U2, PMC with seven doses nearly averted as many cases as RTS,S, while the combination of both was more impactful than either intervention alone. When operational coverage, as seen in Southern Nigeria, increased to a hypothetical target of 80%, cases were reduced beyond the relative increase in coverage. CONCLUSIONS: PMC can substantially reduce clinical and severe cases in the first two years of life in areas with high malaria burden and perennial transmission. A better understanding of the malaria risk profile by age in early childhood and on feasible coverage by age, is needed for selecting an appropriate PMC schedule in a given setting.

Application of mathematical modelling to inform national malaria intervention planning in Nigeria.

BACKGROUND: For their 2021-2025 National Malaria Strategic Plan (NMSP), Nigeria's National Malaria Elimination Programme (NMEP), in partnership with the World Health Organization (WHO), developed a targeted approach to intervention deployment at the local government area (LGA) level as part of the High Burden to High Impact response. Mathematical models of malaria transmission were used to predict the impact of proposed intervention strategies on malaria burden. METHODS: An agent-based model of Plasmodium falciparum transmission was used to simulate malaria morbidity and mortality in Nigeria's 774 LGAs under four possible intervention strategies from 2020 to 2030. The scenarios represented the previously implemented plan (business-as-usual), the NMSP at an 80% or higher coverage level and two prioritized plans according to the resources available to Nigeria. LGAs were clustered into 22 epidemiological archetypes using monthly rainfall, temperature suitability index, vector abundance, pre-2010 parasite prevalence, and pre-2010 vector control coverage. Routine incidence data were used to parameterize seasonality in each archetype. Each LGA's baseline malaria transmission intensity was calibrated to parasite prevalence in children under the age of five years measured in the 2010 Malaria Indicator Survey (MIS). Intervention coverage in the 2010-2019 period was obtained from the Demographic and Health Survey, MIS, the NMEP, and post-campaign surveys. RESULTS: Pursuing a business-as-usual strategy was projected to result in a 5% and 9% increase in malaria incidence in 2025 and 2030 compared with 2020, while deaths were projected to remain unchanged by 2030. The greatest intervention impact was associated with the NMSP scenario with 80% or greater coverage of standard interventions coupled with intermittent preventive treatment in infants and extension of seasonal malaria chemoprevention (SMC) to 404 LGAs, compared to 80 LGAs in 2019. The budget-prioritized scenario with SMC expansion to 310 LGAs, high bed net coverage with new formulations, and increase in effective case management rate at the same pace as historical levels was adopted as an adequate alternative for the resources available. CONCLUSIONS: Dynamical models can be applied for relative assessment of the impact of intervention scenarios but improved subnational data collection systems are required to allow increased confidence in predictions at sub-national level.

Sub-national tailoring of malaria interventions in Mainland Tanzania: simulation of the impact of strata-specific intervention combinations using modelling.

BACKGROUND: To accelerate progress against malaria in high burden countries, a strategic reorientation of resources at the sub-national level is needed. This paper describes how mathematical modelling was used in mainland Tanzania to support the strategic revision that followed the mid-term review of the 2015-2020 national malaria strategic plan (NMSP) and the epidemiological risk stratification at the council level in 2018. METHODS: Intervention mixes, selected by the National Malaria Control Programme, were simulated for each malaria risk strata per council. Intervention mixes included combinations of insecticide-treated bed nets (ITN), indoor residual spraying, larval source management, and intermittent preventive therapies for school children (IPTsc). Effective case management was either based on estimates from the malaria indicator survey in 2016 or set to a hypothetical target of 85%. A previously calibrated mathematical model in OpenMalaria was used to compare intervention impact predictions for prevalence and incidence between 2016 and 2020, or 2022. RESULTS: For each malaria risk stratum four to ten intervention mixes were explored. In the low-risk and urban strata, the scenario without a ITN mass campaign in 2019, predicted high increase in prevalence by 2020 and 2022, while in the very-low strata the target prevalence of less than 1% was maintained at low pre-intervention transmission intensity and high case management. In the moderate and high strata, IPTsc in addition to existing vector control was predicted to reduce the incidence by an additional 15% and prevalence by 22%. In the high-risk strata, all interventions together reached a maximum reduction of 76%, with around 70% of that reduction attributable to high case management and ITNs. Overall, the simulated revised NMSP was predicted to achieve a slightly lower prevalence in 2020 compared to the 2015-2020 NMSP (5.3% vs 6.3%). CONCLUSION: Modelling supported the choice of intervention per malaria risk strata by providing impact comparisons of various alternative intervention mixes to address specific questions relevant to the country. The use of a council-calibrated model, that reproduces local malaria trends, represents a useful tool for compiling available evidence into a single analytical platform, that complement other evidence, to aid national programmes with decision-making processes.

Modeling robust COVID-19 intensive care unit occupancy thresholds for imposing mitigation to prevent exceeding capacities.

In non-pharmaceutical management of COVID-19, occupancy of intensive care units (ICU) is often used as an indicator to inform when to intensify mitigation and thus reduce SARS-CoV-2 transmission, strain on ICUs, and deaths. However, ICU occupancy thresholds at which action should be taken are often selected arbitrarily. We propose a quantitative approach using mathematical modeling to identify ICU occupancy thresholds at which mitigation should be triggered to avoid exceeding the ICU capacity available for COVID-19 patients and demonstrate this approach for the United States city of Chicago. We used a stochastic compartmental model to simulate SARS-CoV-2 transmission and disease progression, including critical cases that would require intensive care. We calibrated the model using daily COVID-19 ICU and hospital census data between March and August 2020. We projected various possible ICU occupancy trajectories from September 2020 to May 2021 with two possible levels of transmission increase and uncertainty in core model parameters. The effect of combined mitigation measures was modeled as a decrease in the transmission rate that took effect when projected ICU occupancy reached a specified threshold. We found that mitigation did not immediately eliminate the risk of exceeding ICU capacity. Delaying action by 7 days increased the probability of exceeding ICU capacity by 10-60% and this increase could not be counteracted by stronger mitigation. Even under modest transmission increase, a threshold occupancy no higher than 60% was required when mitigation reduced the reproductive number Rt to just below 1. At higher transmission increase, a threshold of at most 40% was required with mitigation that reduced Rt below 0.75 within the first two weeks after mitigation. Our analysis demonstrates a quantitative approach for the selection of ICU occupancy thresholds that considers parameter uncertainty and compares relevant mitigation and transmission scenarios. An appropriate threshold will depend on the location, number of ICU beds available for COVID-19, available mitigation options, feasible mitigation strengths, and tolerated durations of intensified mitigation.

Evaluation of different deployment strategies for larviciding to control malaria: a simulation study.

BACKGROUND: Larviciding against malaria vectors in Africa has been limited to indoor residual spraying and insecticide-treated nets, but is increasingly being considered by some countries as a complementary strategy. However, despite progress towards improved larvicides and new tools for mapping or treating mosquito-breeding sites, little is known about the optimal deployment strategies for larviciding in different transmission and seasonality settings. METHODS: A malaria transmission model, OpenMalaria, was used to simulate varying larviciding strategies and their impact on host-seeking mosquito densities, entomological inoculation rate (EIR) and malaria prevalence. Variations in coverage, duration, frequency, and timing of larviciding were simulated for three transmission intensities and four transmission seasonality profiles. Malaria transmission was assumed to follow rainfall with a lag of one month. Theoretical sub-Saharan African settings with Anopheles gambiae as the dominant vector were chosen to explore impact. Relative reduction compared to no larviciding was predicted for each indicator during the simulated larviciding period. RESULTS: Larviciding immediately reduced the predicted host-seeking mosquito densities and EIRs to a maximum that approached or exceeded the simulated coverage. Reduction in prevalence was delayed by approximately one month. The relative reduction in prevalence was up to four times higher at low than high transmission. Reducing larviciding frequency (i.e., from every 5 to 10 days) resulted in substantial loss in effectiveness (54, 45 and 53% loss of impact for host-seeking mosquito densities, EIR and prevalence, respectively). In seasonal settings the most effective timing of larviciding was during or at the beginning of the rainy season and least impactful during the dry season, assuming larviciding deployment for four months. CONCLUSION: The results highlight the critical role of deployment strategies on the impact of larviciding. Overall, larviciding would be more effective in settings with low and seasonal transmission, and at the beginning and during the peak densities of the target species populations. For maximum impact, implementers should consider the practical ranges of coverage, duration, frequency, and timing of larviciding in their respective contexts. More operational data and improved calibration would enable models to become a practical tool to support malaria control programmes in developing larviciding strategies that account for the diversity of contexts.

Geographic and demographic heterogeneity of SARS-CoV-2 diagnostic testing in Illinois, USA, March to December 2020.

BACKGROUND: Availability of SARS-CoV-2 testing in the United States (U.S.) has fluctuated through the course of the COVID-19 pandemic, including in the U.S. state of Illinois. Despite substantial ramp-up in test volume, access to SARS-CoV-2 testing remains limited, heterogeneous, and insufficient to control spread. METHODS: We compared SARS-CoV-2 testing rates across geographic regions, over time, and by demographic characteristics (i.e., age and racial/ethnic groups) in Illinois during March through December 2020. We compared age-matched case fatality ratios and infection fatality ratios through time to estimate the fraction of SARS-CoV-2 infections that have been detected through diagnostic testing. RESULTS: By the end of 2020, initial geographic differences in testing rates had closed substantially. Case fatality ratios were higher in non-Hispanic Black and Hispanic/Latino populations in Illinois relative to non-Hispanic White populations, suggesting that tests were insufficient to accurately capture the true burden of COVID-19 disease in the minority populations during the initial epidemic wave. While testing disparities decreased during 2020, Hispanic/Latino populations consistently remained the least tested at 1.87 tests per 1000 population per day compared with 2.58 and 2.87 for non-Hispanic Black and non-Hispanic White populations, respectively, at the end of 2020. Despite a large expansion in testing since the beginning of the first wave of the epidemic, we estimated that over half (50-80%) of all SARS-CoV-2 infections were not detected by diagnostic testing and continued to evade surveillance. CONCLUSIONS: Systematic methods for identifying relatively under-tested geographic regions and demographic groups may enable policymakers to regularly monitor and evaluate the shifting landscape of diagnostic testing, allowing officials to prioritize allocation of testing resources to reduce disparities in COVID-19 burden and eventually reduce SARS-CoV-2 transmission.

Geographic and demographic heterogeneity of SARS-CoV-2 diagnostic testing in Illinois, USA, March to December 2020.

Background: Availability of SARS-CoV-2 testing in the United States (U.S.) has fluctuated through the course of the COVID-19 pandemic, including in the U.S. state of Illinois. Despite substantial ramp-up in test volume, access to SARS-CoV-2 testing remains limited, heterogeneous, and insufficient to control spread. Methods: We compared SARS-CoV-2 testing rates across geographic regions, over time, and by demographic characteristics (i.e., age and racial/ethnic groups) in Illinois during March through December 2020. We compared age-matched case fatality ratios and infection fatality ratios through time to estimate the fraction of SARS-CoV-2 infections that have been detected through diagnostic testing. Results: By the end of 2020, initial geographic differences in testing rates had closed substantially. Case fatality ratios were higher in non-Hispanic Black and Hispanic/Latino populations in Illinois relative to non-Hispanic White populations, suggesting that tests were insufficient to accurately capture the true burden of COVID-19 disease in the minority populations during the initial epidemic wave. While testing disparities decreased during 2020, Hispanic/Latino populations consistently remained the least tested at 1.87 tests per 1000 population per day compared with 2.58 and 2.87 for non-Hispanic Black and non-Hispanic White populations, respectively, at the end of 2020. Despite a large expansion in testing since the beginning of the first wave of the epidemic, we estimated that over half (50-80%) of all SARS-CoV-2 infections were not detected by diagnostic testing and continued to evade surveillance. Conclusions: Systematic methods for identifying relatively under-tested geographic regions and demographic groups may enable policymakers to regularly monitor and evaluate the shifting landscape of diagnostic testing, allowing officials to prioritize allocation of testing resources to reduce disparities in COVID-19 burden and eventually reduce SARS-CoV-2 transmission.

Beyond national indicators: adapting the Demographic and Health Surveys' sampling strategies and questions to better inform subnational malaria intervention policy.

In malaria-endemic countries, prioritizing intervention deployment to areas that need the most attention is crucial to ensure continued progress. Global and national policy makers increasingly rely on epidemiological data and mathematical modelling to help optimize health decisions at the sub-national level. The Demographic and Health Surveys (DHS) Program is a critical data source for understanding subnational malaria prevalence and intervention coverage, which are used for parameterizing country-specific models of malaria transmission. However, data to estimate indicators at finer resolutions are limited, and surveys questions have a narrow scope. Examples from the Nigeria DHS are used to highlight gaps in the current survey design. Proposals are then made for additional questions and expansions to the DHS and Malaria Indicator Survey sampling strategy that would advance the data analyses and modelled estimates that inform national policy recommendations. Collaboration between the DHS Program, national malaria control programmes, the malaria modelling community, and funders is needed to address the highlighted data challenges.

Identifying the measurements required to estimate rates of COVID-19 transmission, infection, and detection, using variational data assimilation.

We demonstrate the ability of statistical data assimilation (SDA) to identify the measurements required for accurate state and parameter estimation in an epidemiological model for the novel coronavirus disease COVID-19. Our context is an effort to inform policy regarding social behavior, to mitigate strain on hospital capacity. The model unknowns are taken to be: the time-varying transmission rate, the fraction of exposed cases that require hospitalization, and the time-varying detection probabilities of new asymptomatic and symptomatic cases. In simulations, we obtain estimates of undetected (that is, unmeasured) infectious populations, by measuring the detected cases together with the recovered and dead - and without assumed knowledge of the detection rates. Given a noiseless measurement of the recovered population, excellent estimates of all quantities are obtained using a temporal baseline of 101 days, with the exception of the time-varying transmission rate at times prior to the implementation of social distancing. With low noise added to the recovered population, accurate state estimates require a lengthening of the temporal baseline of measurements. Estimates of all parameters are sensitive to the contamination, highlighting the need for accurate and uniform methods of reporting. The aim of this paper is to exemplify the power of SDA to determine what properties of measurements will yield estimates of unknown parameters to a desired precision, in a model with the complexity required to capture important features of the COVID-19 pandemic.

Sub-national stratification of malaria risk in mainland Tanzania: a simplified assembly of survey and routine data.

BACKGROUND: Recent malaria control efforts in mainland Tanzania have led to progressive changes in the prevalence of malaria infection in children, from 18.1% (2008) to 7.3% (2017). As the landscape of malaria transmission changes, a sub-national stratification becomes crucial for optimized cost-effective implementation of interventions. This paper describes the processes, data and outputs of the approach used to produce a simplified, pragmatic malaria risk stratification of 184 councils in mainland Tanzania. METHODS: Assemblies of annual parasite incidence and fever test positivity rate for the period 2016-2017 as well as confirmed malaria incidence and malaria positivity in pregnant women for the period 2015-2017 were obtained from routine district health information software. In addition, parasite prevalence in school children (PfPR5to16) were obtained from the two latest biennial council representative school malaria parasitaemia surveys, 2014-2015 and 2017. The PfPR5to16 served as a guide to set appropriate cut-offs for the other indicators. For each indicator, the maximum value from the past 3 years was used to allocate councils to one of four risk groups: very low (< 1%PfPR5to16), low (1- < 5%PfPR5to16), moderate (5- < 30%PfPR5to16) and high (≥ 30%PfPR5to16). Scores were assigned to each risk group per indicator per council and the total score was used to determine the overall risk strata of all councils. RESULTS: Out of 184 councils, 28 were in the very low stratum (12% of the population), 34 in the low stratum (28% of population), 49 in the moderate stratum (23% of population) and 73 in the high stratum (37% of population). Geographically, most of the councils in the low and very low strata were situated in the central corridor running from the north-east to south-west parts of the country, whilst the areas in the moderate to high strata were situated in the north-west and south-east regions. CONCLUSION: A stratification approach based on multiple routine and survey malaria information was developed. This pragmatic approach can be rapidly reproduced without the use of sophisticated statistical methods, hence, lies within the scope of national malaria programmes across Africa.

Applied mathematical modelling to inform national malaria policies, strategies and operations in Tanzania.

BACKGROUND: More than ever, it is crucial to make the best use of existing country data, and analytical tools for developing malaria control strategies as the heterogeneity in malaria risk within countries is increasing, and the available malaria control tools are expanding while large funding gaps exist. Global and local policymakers, as well as funders, increasingly recognize the value of mathematical modelling as a strategic tool to support decision making. This case study article describes the long-term use of modelling in close collaboration with the National Malaria Control Programme (NMCP) in Tanzania, the challenges encountered and lessons learned. CASE DESCRIPTION: In Tanzania, a recent rebound in prevalence led to the revision of the national malaria strategic plan with interventions targeted to the malaria risk at the sub-regional level. As part of the revision, a mathematical malaria modelling framework for setting specific predictions was developed and used between 2016 and 2019 to (1) reproduce setting specific historical malaria trends, and (2) to simulate in silico the impact of future interventions. Throughout the project, multiple stakeholder workshops were attended and the use of mathematical modelling interactively discussed. EVALUATION: In Tanzania, the model application created an interdisciplinary and multisectoral dialogue platform between modellers, NMCP and partners and contributed to the revision of the national malaria strategic plan by simulating strategies suggested by the NMCP. The uptake of the modelling outputs and sustained interest by the NMCP were critically associated with following factors: (1) effective sensitization to the NMCP, (2) regular and intense communication, (3) invitation for the modellers to participate in the strategic plan process, and (4) model application tailored to the local context. CONCLUSION: Empirical data analysis and its use for strategic thinking remain the cornerstone for evidence-based decision-making. Mathematical impact modelling can support the process both by unifying all stakeholders in one strategic process and by adding new key evidence required for optimized decision-making. However, without a long-standing partnership, it will be much more challenging to sensibilize programmes to the usefulness and sustained use of modelling and local resources within the programme or collaborating research institutions need to be mobilized.

Simulating the council-specific impact of anti-malaria interventions: A tool to support malaria strategic planning in Tanzania.

INTRODUCTION: The decision-making process for malaria control and elimination strategies has become more challenging. Interventions need to be targeted at council level to allow for changing malaria epidemiology and an increase in the number of possible interventions. Models of malaria dynamics can support this process by simulating potential impacts of multiple interventions in different settings and determining appropriate packages of interventions for meeting specific expected targets. METHODS: The OpenMalaria model of malaria dynamics was calibrated for all 184 councils in mainland Tanzania using data from malaria indicator surveys, school parasitaemia surveys, entomological surveillance, and vector control deployment data. The simulations were run for different transmission intensities per region and five interventions, currently or potentially included in the National Malaria Strategic Plan, individually and in combination. The simulated prevalences were fitted to council specific prevalences derived from geostatistical models to obtain council specific predictions of the prevalence and number of cases between 2017 and 2020. The predictions were used to evaluate in silico the feasibility of the national target of reaching a prevalence of below 1% by 2020, and to suggest alternative intervention stratifications for the country. RESULTS: The historical prevalence trend was fitted for each council with an agreement of 87% in 2016 (95%CI: 0.84-0.90) and an agreement of 90% for the historical trend (2003-2016) (95%CI: 0.87-0.93) The current national malaria strategy was expected to reduce the malaria prevalence between 2016 and 2020 on average by 23.8% (95% CI: 19.7%-27.9%) if current case management levels were maintained, and by 52.1% (95% CI: 48.8%-55.3%) if the case management were improved. Insecticide treated nets and case management were the most cost-effective interventions, expected to reduce the prevalence by 25.0% (95% CI: 19.7%-30.2) and to avert 37 million cases between 2017 and 2020. Mass drug administration was included in most councils in the stratification selected for meeting the national target at minimal costs, expected to reduce the prevalence by 77.5% (95%CI: 70.5%-84.5%) and to avert 102 million cases, with almost twice higher costs than those of the current national strategy. In summary, the model suggested that current interventions are not sufficient to reach the national aim of a prevalence of less than 1% by 2020 and a revised strategic plan needs to consider additional, more effective interventions, especially in high transmission areas and that the targets need to be revisited. CONCLUSION: The methodology reported here is based on intensive interactions with the NMCP and provides a helpful tool for assessing the feasibility of country specific targets and for determining which intervention stratifications at sub-national level will have most impact. This country-led application could support strategic planning of malaria control in many other malaria endemic countries.

The potential of pregnant women as a sentinel population for malaria surveillance.

BACKGROUND: With increasing spatial heterogeneity of malaria transmission and a shift of the disease burden towards older children and adults, pregnant women attending antenatal care (ANC) have been proposed as a pragmatic sentinel population for malaria surveillance. However, the representativeness of routine ANC malaria test-positivity and its relationship with prevalence in other population subgroups are yet to be investigated. METHODS: Monthly ANC malaria test-positivity data from all Tanzanian health facilities for January 2014 to May 2016 was compared to prevalence data from the School Malaria Parasitaemia Survey 2015, the Malaria Indicator Survey (MIS) 2015/16, the Malaria Atlas Project 2015, and a Bayesian model fitted to MIS data. Linear regression was used to describe the difference between malaria test-positivity in pregnant women and respective comparison groups as a function of ANC test-positivity and potential covariates. RESULTS: The relationship between ANC test-positivity and survey prevalence in children follows spatially and biologically meaningful patterns. However, the uncertainty of the relationship was substantial, particularly in areas with high or perennial transmission. In comparison, modelled data estimated higher prevalence in children at low transmission intensities and lower prevalence at higher transmission intensities. CONCLUSIONS: Pregnant women attending ANC are a pragmatic sentinel population to assess heterogeneity and trends in malaria prevalence in Tanzania. Yet, since ANC malaria test-positivity cannot be used to directly predict the prevalence in other population subgroups, complementary community-level measurements remain highly relevant.

Nationwide school malaria parasitaemia survey in public primary schools, the United Republic of Tanzania.

BACKGROUND: A nationwide, school, malaria survey was implemented to assess the risk factors of malaria prevalence and bed net use among primary school children in mainland Tanzania. This allowed the mapping of malaria prevalence at council level and assessment of malaria risk factors among school children. METHODS: A cross-sectional, school, malaria parasitaemia survey was conducted in 25 regions, 166 councils and 357 schools in three phases: (1) August to September 2014; (2) May 2015; and, (3) October 2015. Children were tested for malaria parasites using rapid diagnostic tests and were interviewed about household information, parents' education, bed net indicators as well as recent history of fever. Multilevel mixed effects logistic regression models were fitted to estimate odds ratios of risk factors for malaria infection and for bed net use while adjusting for school effect. RESULTS: In total, 49,113 children were interviewed and tested for malaria infection. The overall prevalence of malaria was 21.6%, ranging from < 0.1 to 53% among regions and from 0 to 76.4% among councils. The malaria prevalence was below 5% in 62 of the 166 councils and above 50% in 18 councils and between 5 and 50% in the other councils. The variation of malaria prevalence between schools was greatest in regions with a high mean prevalence, while the variation was marked by a few outlying schools in regions with a low mean prevalence. Overall, 70% of the children reported using mosquito nets, with the highest percentage observed among educated parents (80.7%), low land areas (82.7%) and those living in urban areas (82.2%). CONCLUSIONS: The observed prevalence among school children showed marked variation at regional and sub-regional levels across the country. Findings of this survey are useful for updating the malaria epidemiological profile and for stratification of malaria transmission by region, council and age groups, which is essential for guiding resource allocation, evaluation and prioritization of malaria interventions.