The past matters: estimating intrinsic hookworm transmission intensity in areas with past mass drug administration to control lymphatic filariasis

Forecasting the effectiveness of the DeWorm3 trial in interrupting the transmission of soil-transmitted helminths in three study sites in Benin, India and Malawi

Background

The DeWorm3 project is an ongoing cluster-randomised trial assessing the feasibility of interrupting the transmission of soil-transmitted helminths (STH) through mass drug administration (MDA) using study sites in India, Malawi and Benin. In this article, we describe an approach which uses a combination of statistical and mathematical methods to forecast the outcome of the trial with respect to its stated goal of reducing the prevalence of infection to below 2%.

Methods

Our approach is first to define the local patterns of transmission within each study site, which is achieved by statistical inference of key epidemiological parameters using the baseline epidemiological measures of age-related prevalence and intensity of STH infection which have been collected by the DeWorm3 trials team. We use these inferred parameters to calibrate an individual-based stochastic simulation of the trial at the cluster and study site level, which is subsequently run to forecast the future prevalence of STH infections. The simulator takes into account both the uncertainties in parameter estimation and the variability inherent in epidemiological and demographic processes in the simulator. We interpret the forecast results from our simulation with reference to the stated goal of the DeWorm3 trial, to achieve a target of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\le 2\%$$\end{document}≤2% prevalence at a point 24 months post-cessation of MDA.

Results

Simulated output predicts that the two arms will be distinguishable from each other in all three country sites at the study end point. In India and Malawi, measured prevalence in the intervention arm is below the threshold with a high probability (90% and 95%, respectively), but in Benin the heterogeneity between clusters prevents the arm prevalence from being reduced below the threshold value. At the level of individual study arms within each site, heterogeneity among clusters leads to a very low probability of achieving complete elimination in an intervention arm, yielding a post-study scenario with widespread elimination but a few ‘hot spot’ areas of persisting STH transmission.

Conclusions

Our results suggest that geographical heterogeneities in transmission intensity and worm aggregation have a large impact on the effect of MDA. It is important to accurately assess cluster-level, or even smaller scale, heterogeneities in factors which influence transmission and aggregation for a clearer perspective on projecting the outcomes of MDA control of STH and other neglected tropical diseases.

Baseline patterns of infection in regions of Benin, Malawi and India seeking to interrupt transmission of soil transmitted helminths (STH) in the DeWorm3 trial

Global efforts to control morbidity associated with soil-transmitted helminth infections (STH) have focused largely on the targeted treatment of high-risk groups, including children and pregnant women. However, it is not clear when such programs can be discontinued and there are concerns about the sustainability of current STH control programs. The DeWorm3 project is a large multi-country community cluster randomized trial in Benin, India and Malawi designed to determine the feasibility of interrupting the transmission of STH using community-wide delivery of mass drug administration (MDA) with anthelmintics over multiple rounds. Here, we present baseline data and estimate key epidemiological parameters important in determining the likelihood of transmission interruption in the DeWorm3 trial. A baseline census was conducted in October-December 2017 in India, November-December 2017 in Malawi and in January-February 2018 in Benin. The baseline census enumerated all members of each household and collected demographic data and information on occupation, assets, and access to water, sanitation and hygiene (WASH). Each study site was divided into 40 clusters of at least 1,650 individuals per cluster. Clusters were randomized to receive twice yearly community-wide MDA with albendazole (GSK) targeting eligible individuals of all ages (20 clusters), or to receive the standard-of-care deworming program targeting children provided in each country. In each site, a randomly selected group of 150 individuals per cluster (6,000 total per site) was selected from the baseline census using stratified random sampling, and each individual provided a single stool sample for analysis of STH infection using the Kato-Katz technique. Study site, household and individual characteristics were summarized as appropriate. We estimated key epidemiological parameters including the force of infection and the degree of parasite aggregation within the population. The DeWorm3 sites range in population from 94,969 to 140,932. The population age distribution varied significantly by site, with the highest proportion of infants and young children in Malawi and the highest proportion of adults in India. The baseline age- and cluster-weighted prevalence, as measured by Kato-Katz, varied across sites and by species, Baseline hookworm prevalence in India was 21.4% (95% CI: 20.4-22.4%), while prevalence of Ascaris and Trichuris by Kato-Katz was low (0.1% and 0.3% overall). In Malawi, the overall age- and cluster-weighted STH prevalence was 7.7% (95% CI: 7.1-8.4%) predominantly driven by hookworm infections (7.4%) while Ascaris (0.1%) and Trichuris (0.3%) infections were rare. In Benin, the overall age- and cluster-weighted prevalence was significantly lower (5.6%, 95% CI: 5.1-6.2%) and Ascaris (2.0%, 95% CI: 1.6-2.3%) was more common than in other sites. Ascaris infections were more likely to be moderate- or heavy-intensity (43.7%, unweighted) compared to hookworm (5.0%). The force of infection for hookworm was highest in adults in India and Malawi but appeared relatively stable across age groups in Benin. These data demonstrate the significant variability between the sites in terms of demography, socio-economic status and environmental characteristics. In addition, the baseline prevalence and intensity data from DeWorm3 suggest that each site has unique epidemiologic characteristics that will be critical in determining correlates of achieving STH transmission interruption in the DeWorm3 trial. Trial registration: The trial was registered at ClinicalTrials.gov (NCT03014167).

Impact of annual and semi-annual mass drug administration for Lymphatic Filariasis and Onchocerciasis on Hookworm Infection in Côte d'Ivoire

Mass Drug Administration (MDA) programs to eliminate Lymphatic Filariasis (LF) in western Africa use the anthelminthics ivermectin plus albendazole. These drugs have the potential to impact also Soil-Transmitted Helminth (STH) infections, since the drugs have a broad range of anthelminthic activity. Integration of preventive chemotherapy efforts for LF, onchocerciasis and STH is recommended by the World Health Organization (WHO) in order to avoid duplication of MDA and to reduce costs. The objective of the current study was to determine whether five semi-annual rounds of community-wide MDA to eliminate LF and onchocerciasis have a greater impact on STH than three annual rounds of MDA with similar compliance. The effects of MDA using ivermectin (IVM, 0.2 mg/kg) combined with albendazole (ALB, 400 mg) on the prevalence and intensity of hookworm infection were evaluated in the Abengourou (annual MDA) and Akoupé (semi-annual MDA) health Districts in eastern Côte d’Ivoire from 2014 to 2017. A cross-sectional approach was used together with mixed logistic regression, and mixed linear models. Subjects were tested for STH using the Kato-Katz technique before the first round of MDA and 12, 24, and 36 months after the first round of MDA. The mean self-reported MDA compliance assessed during the survey was 65%, and no difference was observed between treatment areas. These results were confirmed by an independent coverage survey as recommended by WHO. Hookworm was the most prevalent STH species in both areas (23.9% vs 12.4%) and the prevalence of other STH species was less than 1%. The crude prevalence of hookworm dropped significantly, from 23.9% to 5.5% (p <0.001, 77% reduction) in the annual MDA treatment area and from 12.4% to 1.9% (p <0.001, 85% reduction) in the semi-annual treatment area. The average intensity of hookworm infection decreased in the annual MDA area (406.2 epg to 118.3 epg), but not in the semi-annual MDA area (804.9 epg to 875.0 epg). Moderate and heavy infections (1% and 1.3% at baseline) were reduced to 0% and 0.4% in the annual and semi-annual treatment areas, respectively. Using a mixed logistic regression model, and after adjusting for baseline prevalence, only the year 2 re-examination showed a difference in prevalence between treatments (OR: 2.26 [95% CI: 1.03, 4.98], p = 0.043). Analysis of intensity of hookworm infection indicated also that treatment differences varied by follow-up visit. In conclusion twelve months after the last treatment cycle, three annual and five semi-annual rounds of community-wide MDA with the combination of IVM and ALB showed strong, but similar impact on hookworm prevalence and intensity in eastern Côte d’Ivoire. Therefore, an annual MDA regimen seems to be an efficient strategy to control hookworm infection in endemic areas with low and moderate infection prevalence.

Trial registration: The study was registered at ClinicalTrial.gov under the number NTC02032043.

Community-wide MDA to eliminate LF and onchocerciasis has the beneficial effect to reduce also STH infections. The objective of the current study was to determine whether five semi-annual rounds of MDA have a greater impact on STH than three annual rounds of MDA using ivermectin combined with albendazole. In Abengourou and Akoupé health Districts in eastern Côte d’Ivoire the prevalence and intensity of hookworm infection were evaluated before and after MDA. Prior to MDA and after each annual treatment cycle, study participants were tested for STH using the Kato-Katz technique. The mean MDA compliance assessed during the survey was 65%, and no difference was observed between treatment areas. Compliance results were confirmed by an independent coverage survey as recommended by WHO. Hookworm was the most prevalent STH species in both areas and the prevalence of other STH species was less than 1%. The crude prevalence of hookworm dropped significantly, from 23.9% to 5.5% in the annual and from 12.4% to 1.9% in the semi-annual treatment areas. The intensity of hookworm infection in infected persons decreased significantly in the annual MDA area (406.2 epg to 118.3 epg, p = 0.017), but not in the semi-annual MDA area (804.9 epg to 875.0 epg, p = 0.216). Moderate and heavy infections were reduced to less than 1% in both treatment areas. Three annual and five semi-annual rounds of community-wide MDA with the combination of IVM and ALB showed strong, but similar effects on hookworm prevalence and intensity in eastern Côte d’Ivoire. Therefore, an annual MDA regimen seems to be an efficient strategy for controlling hookworm infection in endemic areas with low and moderate infection prevalence.

Are We on Our Way to Achieving the 2020 Goals for Schistosomiasis Morbidity Control Using Current World Health Organization Guidelines?

Background

Schistosomiasis remains an endemic parasitic disease affecting millions of people around the world. The World Health Organization (WHO) has set goals of controlling morbidity to be reached by 2020, along with elimination as a public health problem in certain regions by 2025. Mathematical models of parasite transmission and treatment impact have been developed to assist in controlling the morbidity caused by schistosomiasis. These models can inform and guide implementation policy for mass drug administration programs, and help design monitoring and evaluation activities.

Methods

We use these models to predict whether the guidelines set by the WHO are on track for achieving their 2020 goal for the control of morbidity, specifically for Schistosoma mansoni. We examine whether programmatic adaptations; namely increases in treatment coverage and/or expansion to adult inclusion in treatment, will improve the likelihood of reaching the WHO goals.

Results

We find that in low-prevalence settings, the goals are likely to be attainable under current WHO guidelines, but in moderate to high-prevalence settings, the goals are less likely to be achieved unless treatment coverage is increased and expanded to at least 85% for school-aged children and 40% for adults.

Conclusions

To improve the likelihood of reaching the WHO goals, programmatic adaptations are required, particularly for moderate- to high-prevalence settings. Furthermore, improvements in adherence to treatment, potential development of candidate vaccines, and enhanced snail control and WASH (water, sanitation, and hygiene) measures will all assist in achieving the goals.

Heterogeneity in transmission parameters of hookworm infection within the baseline data from the TUMIKIA study in Kenya

Background

As many countries with endemic soil-transmitted helminth (STH) burdens achieve high coverage levels of mass drug administration (MDA) to treat school-aged and pre-school-aged children, understanding the detailed effects of MDA on the epidemiology of STH infections is desirable in formulating future policies for morbidity and/or transmission control. Prevalence and mean intensity of infection are characterized by heterogeneity across a region, leading to uncertainty in the impact of MDA strategies. In this paper, we analyze this heterogeneity in terms of factors that govern the transmission dynamics of the parasite in the host population.

Results

Using data from the TUMIKIA study in Kenya (cluster STH prevalence range at baseline: 0–63%), we estimated these parameters and their variability across 120 population clusters in the study region, using a simple parasite transmission model and Gibbs-sampling Monte Carlo Markov chain techniques. We observed great heterogeneity in R₀ values, with estimates ranging from 1.23 to 3.27, while k-values (which vary inversely with the degree of parasite aggregation within the human host population) range from 0.007 to 0.29 in a positive association with increasing prevalence. The main finding of this study is the increasing trend for greater parasite aggregation as prevalence declines to low levels, reflected in the low values of the negative binomial parameter k in clusters with low hookworm prevalence. Localized climatic and socioeconomic factors are investigated as potential drivers of these observed epidemiological patterns.

Conclusions

Our results show that lower prevalence is associated with higher degrees of aggregation and hence prevalence alone is not a good indicator of transmission intensity. As a consequence, approaches to MDA and monitoring and evaluation of community infection status may need to be adapted as transmission elimination is aimed for by targeted treatment approaches.

Community-level epidemiology of soil-transmitted helminths in the context of school-based deworming: Baseline results of a cluster randomised trial on the coast of Kenya

TRIAL REGISTRATION

ClinicalTrials.gov NCT02397772.

The design of schistosomiasis monitoring and evaluation programmes: The importance of collecting adult data to inform treatment strategies for Schistosoma mansoni

Monitoring and evaluation (M&E) programmes are used to collect data which are required to assess the impact of current interventions on their progress towards achieving the World Health Organization (WHO) goals of morbidity control and elimination as a public health problem for schistosomiasis. Prevalence and intensity of infection data are typically collected from school-aged children (SAC) as they are relatively easy to sample and are thought to be most likely to be infected by schistosome parasites. However, adults are also likely to be infected. We use three different age-intensity profiles of infection for Schistosoma mansoni with low, moderate and high burdens of infection in adults to investigate how the age distribution of infection impacts the mathematical model generated recommendations of the preventive chemotherapy coverage levels required to achieve the WHO goals. We find that for moderate prevalence regions, regardless of the burden of infection in adults, treating SAC only may achieve the WHO goals. However, for high prevalence regions with a high burden of infection in adults, adult treatment is required to meet the WHO goals. Hence, we show that the optimal treatment strategy for a defined region requires consideration of the burden of infection in adults as it cannot be based solely on the prevalence of infection in SAC. Although past epidemiological data have informed mathematical models for the transmission and control of schistosome infections, more accurate and detailed data are required from M&E programmes to accurately determine the optimal treatment strategy for a defined region. We highlight the importance of collecting prevalence and intensity of infection data from a broader age-range, specifically the inclusion of adult data at baseline (prior to treatment) and throughout the treatment programme if possible, rather than SAC only, to accurately determine the treatment strategy for a defined region. Furthermore, we discuss additional epidemiological data, such as individual longitudinal adherence to treatment, that should ideally be collected in M&E programmes.

Schistosomiasis remains an endemic parasitic disease affecting millions of people around the world. The World Health Organization (WHO) has set goals of morbidity control and elimination as a public health problem for schistosomiasis defined by reaching ≤5% and ≤1% prevalence of heavy-intensity infections in school-aged children, respectively. Monitoring and evaluation (M&E) programmes are used to collect data which can inform treatment strategies required in a defined area and can also aid in assessing the progress of implemented treatment strategies. Due to programmatic and financial constraints, M&E data are typically collected from school-aged children as they are thought to be most likely to be infected. We highlight that adults should be included within M&E programmes by showing how the burden of infection in adults impacts our mathematical model recommendations of the treatment coverage levels required to reach the WHO goals for schistosomiasis. Our results highlight the importance of collecting data from a broader age-range, specifically the inclusion of adult data at baseline (prior to treatment) and throughout the treatment programme if possible. Improving M&E programmes to incorporate collection of such data will allow for more accurate determination of the optimal treatment strategy for a defined region.

Defining stopping criteria for ending randomized clinical trials that investigate the interruption of transmission of soil-transmitted helminths employing mass drug administration

The current World Health Organization strategy to address soil-transmitted helminth (STH) infections in children is based on morbidity control through routine deworming of school and pre-school aged children. However, given that transmission continues to occur as a result of persistent reservoirs of infection in untreated individuals (including adults) and in the environment, in many settings such a strategy will need to be continued for very extended periods of time, or until social, economic and environmental conditions result in interruption of transmission. As a result, there is currently much discussion surrounding the possibility of accelerating the interruption of transmission using alternative strategies of mass drug administration (MDA). However, the feasibility of achieving transmission interruption using MDA remains uncertain due to challenges in sustaining high MDA coverage levels across entire communities. The DeWorm3 trial, designed to test the feasibility of interrupting STH transmission, is currently ongoing. In DeWorm3, three years of high treatment coverage-indicated by mathematical models as necessary for breaking transmission-will be followed by two years of surveillance. Given the fast reinfection (bounce-back) rates of STH, a two year no treatment period is regarded as adequate to assess whether bounce-back or transmission interruption have occurred in a given location. In this study, we investigate if criteria to determine whether transmission interruption is unlikely can be defined at earlier timepoints. A stochastic, individual-based simulation model is employed to simulate core aspects of the DeWorm3 community-based cluster-randomized trial. This trial compares a control arm (annual treatment of children alone with MDA) with an intervention arm (community-wide biannual treatment with MDA). Simulations were run for each scenario for both Ascaris lumbricoides and hookworm (Necator americanus). A range of threshold prevalences measured at six months after the last round of MDA and the impact of MDA coverage levels were evaluated to see if the likelihood of bounce-back or elimination could reliably be assessed at that point, rather than after two years of subsequent surveillance. The analyses suggest that all clusters should be assessed for transmission interruption after two years of surveillance, unless transmission interruption can be effectively ruled out through evidence of low treatment coverage. Models suggest a tight range of homogenous prevalence estimates following high coverage MDA across clusters which do not allow for discrimination between bounce back or transmission interruption within 24 months following cessation of MDA.

Testing for soil-transmitted helminth transmission elimination: Analysing the impact of the sensitivity of different diagnostic tools

In recent years, an increased focus has been placed upon the possibility of the elimination of soil-transmitted helminth (STH) transmission using various interventions including mass drug administration. The primary diagnostic tool recommended by the WHO is the detection of STH eggs in stool using the Kato-Katz (KK) method. However, detecting infected individuals using this method becomes increasingly difficult as the intensity of infection decreases. Newer techniques, such as qPCR, have been shown to have greater sensitivity than KK, especially at low prevalence. However, the impact of using qPCR on elimination thresholds is yet to be investigated. In this paper, we aim to quantify how the sensitivity of these two diagnostic tools affects the optimal prevalence threshold at which to declare the interruption of transmission with a defined level of confidence. A stochastic, individual-based STH transmission model was used in this study to simulate the transmission dynamics of Ascaris and hookworm. Data from a Kenyan deworming study were used to parameterize the diagnostic model which was based on egg detection probabilities. The positive and negative predictive values (PPV and NPV) were calculated to assess the quality of any given threshold, with the optimal threshold value taken to be that at which both were maximised. The threshold prevalence of infection values for declaring elimination of Ascaris transmission were 6% and 12% for KK and qPCR respectively. For hookworm, these threshold values are lower at 0.5% and 2% respectively. Diagnostic tests with greater sensitivity are becoming increasingly important as we approach the elimination of STH transmission in some regions of the world. For declaring the elimination of transmission, using qPCR to diagnose STH infection results in the definition of a higher prevalence, than when KK is used.

Prospects for elimination of soil-transmitted helminths

PURPOSE OF REVIEW

Soil-transmitted helminths (STH) are endemic in 120 countries and are associated with substantial morbidity and loss of economic productivity. Although current WHO guidelines focus on morbidity control through mass drug administration (MDA), there is global interest in whether a strategy targeting disease elimination might be feasible in some settings. This review summarizes the prospects for switching from control to an elimination strategy.

RECENT FINDINGS

STH control efforts have reduced the intensity of infections in targeted populations with associated reductions in morbidity. However, adults are not frequently targeted and remain important reservoirs for reinfection of treated children. Recent modeling suggests that transmission interruption may be possible through expanded community-wide delivery of MDA, the feasibility of which has been demonstrated by other programs. However, these models suggest that high levels of coverage and compliance must be achieved. Potential challenges include the risk of prematurely dismantling STH programs and the potential increased risk of antihelminthic resistance.

SUMMARY

Elimination of STH may offer an opportunity to eliminate substantial STH-related morbidity while reducing resource needs of neglected tropical disease programs. Evidence from large community trials is needed to determine the feasibility of interrupting the transmission of STH in some geographic settings.

Identifying optimal threshold statistics for elimination of hookworm using a stochastic simulation model

BACKGROUND

There is an increased focus on whether mass drug administration (MDA) programmes alone can interrupt the transmission of soil-transmitted helminths (STH). Mathematical models can be used to model these interventions and are increasingly being implemented to inform investigators about expected trial outcome and the choice of optimum study design. One key factor is the choice of threshold for detecting elimination. However, there are currently no thresholds defined for STH regarding breaking transmission.

METHODS

We develop a simulation of an elimination study, based on the DeWorm3 project, using an individual-based stochastic disease transmission model in conjunction with models of MDA, sampling, diagnostics and the construction of study clusters. The simulation is then used to analyse the relationship between the study end-point elimination threshold and whether elimination is achieved in the long term within the model. We analyse the quality of a range of statistics in terms of the positive predictive values (PPV) and how they depend on a range of covariates, including threshold values, baseline prevalence, measurement time point and how clusters are constructed.

RESULTS

End-point infection prevalence performs well in discriminating between villages that achieve interruption of transmission and those that do not, although the quality of the threshold is sensitive to baseline prevalence and threshold value. Optimal post-treatment prevalence threshold value for determining elimination is in the range 2% or less when the baseline prevalence range is broad. For multiple clusters of communities, both the probability of elimination and the ability of thresholds to detect it are strongly dependent on the size of the cluster and the size distribution of the constituent communities. Number of communities in a cluster is a key indicator of probability of elimination and PPV. Extending the time, post-study endpoint, at which the threshold statistic is measured improves PPV value in discriminating between eliminating clusters and those that bounce back.

CONCLUSIONS

The probability of elimination and PPV are very sensitive to baseline prevalence for individual communities. However, most studies and programmes are constructed on the basis of clusters. Since elimination occurs within smaller population sub-units, the construction of clusters introduces new sensitivities for elimination threshold values to cluster size and the underlying population structure. Study simulation offers an opportunity to investigate key sources of sensitivity for elimination studies and programme designs in advance and to tailor interventions to prevailing local or national conditions.