A Sequential Cross-Sectional Analysis Producing Robust Weekly COVID-19 Rates for South East Asian Countries

The COVID-19 pandemic has expanded fast over the world, affecting millions of people and generating serious health, social, and economic consequences. All South East Asian countries have experienced the pandemic, with various degrees of intensity and response. As the pandemic progresses, it is important to track and analyse disease trends and patterns to guide public health policy and treatments. In this paper, we carry out a sequential cross-sectional study to produce reliable weekly COVID-19 death (out of cases) rates for South East Asian countries for the calendar years 2020, 2021, and 2022. The main objectives of this study are to characterise the trends and patterns of COVID-19 death rates in South East Asian countries through time, as well as compare COVID-19 rates among countries and regions in South East Asia. Our raw data are (daily) case and death counts acquired from "Our World in Data", which, however, for some countries and time periods, suffer from sparsity (zero or small counts), and therefore require a modelling approach where information is adaptively borrowed from the overall dataset where required. Therefore, a sequential cross-sectional design will be utilised, that will involve examining the data week by week, across all countries. Methodologically, this is achieved through a two-stage random effect shrinkage approach, with estimation facilitated by nonparametric maximum likelihood.

Estimating dengue transmission intensity from serological data: A comparative analysis using mixture and catalytic models

Background

Dengue virus (DENV) infection is a global health concern of increasing magnitude. To target intervention strategies, accurate estimates of the force of infection (FOI) are necessary. Catalytic models have been widely used to estimate DENV FOI and rely on a binary classification of serostatus as seropositive or seronegative, according to pre-defined antibody thresholds. Previous work has demonstrated the use of thresholds can cause serostatus misclassification and biased estimates. In contrast, mixture models do not rely on thresholds and use the full distribution of antibody titres. To date, there has been limited application of mixture models to estimate DENV FOI.

Methods

We compare the application of mixture models and time-constant and time-varying catalytic models to simulated data and to serological data collected in Vietnam from 2004 to 2009 (N ≥ 2178) and Indonesia in 2014 (N = 3194).

Results

The simulation study showed larger mean FOI estimate bias from the time-constant and time-varying catalytic models (-0.007 (95% Confidence Interval (CI): -0.069, 0.029) and -0.006 (95% CI -0.095, 0.043)) than from the mixture model (0.001 (95% CI -0.036, 0.065)). Coverage of the true FOI was > 95% for estimates from both the time-varying catalytic and mixture model, however the latter had reduced uncertainty. When applied to real data from Vietnam, the mixture model frequently produced higher FOI and seroprevalence estimates than the catalytic models.

Conclusions

Our results suggest mixture models represent valid, potentially less biased, alternatives to catalytic models, which could be particularly useful when estimating FOI from data with largely overlapping antibody titre distributions.

Characterising the transmission intensity of dengue virus is essential to inform the implementation of interventions, such as vector control and vaccination, and to better understand the environmental drivers of transmission locally and globally. It is therefore important to understand how methodological differences and model choice may influence the accuracy of estimates of transmission intensity. Using a simulation study, we assessed the performance of catalytic and mixture models to reconstruct the force of infection (FOI) from simulated antibody titre data. Furthermore, we estimated the FOI of dengue virus from antibody titre data collected in Vietnam and Indonesia. The models produced consistent estimates of FOI when they were applied to data with clear separation between the distributions of seronegative and seropositive antibody titres. We observed greater bias in FOI estimates obtained from catalytic models than from mixture models when they were applied to data with high overlap in the bimodal distribution of antibody titres. Our results indicate that mixture models could be preferential to estimate dengue virus FOI when the antibody titre distributions of the seronegative and seropositive components largely overlap.

Identifying immunity gaps for measles using Belgian serial serology data

Vaccine-preventable diseases, such as measles, have been re-emerging in countries with moderate to high vaccine uptake. It is increasingly important to identify and close immunity gaps and increase coverage of routine childhood vaccinations, including two doses of the measles-mumps-rubella vaccine (MMR). Here, we present a simple cohort model relying on a Bayesian approach to evaluate the evolution of measles seroprevalence in Belgium using the three most recent cross-sectional serological survey data collections (2002, 2006 and 2013) and information regarding vaccine properties. We find measles seroprevalence profiles to be similar for the different regions in Belgium. These profiles exhibit a drop in seroprevalence in birth cohorts that were offered vaccination at suboptimal coverages in the first years after routine vaccination has been started up. This immunity gap is observed across all cross-sectional survey years, although it is more pronounced in survey year 2013. At present, the COVID-19 pandemic could negatively impact the immunization coverage worldwide, thereby increasing the need for additional immunization programs in groups of children that are impacted by this. Therefore, it is now even more important to identify existing immunity gaps and to sustain and reach vaccine-derived measles immunity goals.

A threshold-free approach with age-dependency for estimating malaria seroprevalence

Background

In malaria serology analysis, the standard approach to obtain seroprevalence, i.e the proportion of seropositive individuals in a population, is based on a threshold which is used to classify individuals as seropositive or seronegative. The choice of this threshold is often arbitrary and is based on methods that ignore the age-dependency of the antibody distribution.

Methods

Using cross-sectional antibody data from the Western Kenyan Highlands, this paper introduces a novel approach that has three main advantages over the current threshold-based approach: it avoids the use of thresholds; it accounts for the age dependency of malaria antibodies; and it allows us to propagate the uncertainty from the classification of individuals into seropositive and seronegative when estimating seroprevalence. The reversible catalytic model is used as an example for illustrating how to propagate this uncertainty into the parameter estimates of the model.

Results

This paper finds that accounting for age-dependency leads to a better fit to the data than the standard approach which uses a single threshold across all ages. Additionally, the paper also finds that the proposed threshold-free approach is more robust against the selection of different age-groups when estimating seroprevalence.

Conclusion

The novel threshold-free approach presented in this paper provides a statistically principled and more objective approach to estimating malaria seroprevalence. The introduced statistical framework also provides a means to compare results across studies which may use different age ranges for the estimation of seroprevalence.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12936-021-04022-4.

A unified and flexible modelling framework for the analysis of malaria serology data

Serology data are an increasingly important tool in malaria surveillance, especially in low transmission settings where the estimation of parasite-based indicators is often problematic. Existing methods rely on the use of thresholds to identify seropositive individuals and estimate transmission intensity, while making assumptions about the temporal dynamics of malaria transmission that are rarely questioned. Here, we present a novel threshold-free approach for the analysis of malaria serology data which avoids dichotomization of continuous antibody measurements and allows us to model changes in the antibody distribution across age in a more flexible way. The proposed unified mechanistic model combines the properties of reversible catalytic and antibody acquisition models, and allows for temporally varying boosting and seroconversion rates. Additionally, as an alternative to the unified mechanistic model, we also propose an empirical approach to analysis where modelling of the age-dependency is informed by the data rather than biological assumptions. Using serology data from Western Kenya, we demonstrate both the usefulness and limitations of the novel modelling framework.

Linking longitudinal and cross-sectional biomarker data to understand host-pathogen dynamics: Leptospira in California sea lions (Zalophus californianus) as a case study

Confronted with the challenge of understanding population-level processes, disease ecologists and epidemiologists often simplify quantitative data into distinct physiological states (e.g. susceptible, exposed, infected, recovered). However, data defining these states often fall along a spectrum rather than into clear categories. Hence, the host-pathogen relationship is more accurately defined using quantitative data, often integrating multiple diagnostic measures, just as clinicians do to assess their patients. We use quantitative data on a major neglected tropical disease (Leptospira interrogans) in California sea lions (Zalophus californianus) to improve individual-level and population-level understanding of this Leptospira reservoir system. We create a "host-pathogen space" by mapping multiple biomarkers of infection (e.g. serum antibodies, pathogen DNA) and disease state (e.g. serum chemistry values) from 13 longitudinally sampled, severely ill individuals to characterize changes in these values through time. Data from these individuals describe a clear, unidirectional trajectory of disease and recovery within this host-pathogen space. Remarkably, this trajectory also captures the broad patterns in larger cross-sectional datasets of 1456 wild sea lions in all states of health but sampled only once. Our framework enables us to determine an individual's location in their time-course since initial infection, and to visualize the full range of clinical states and antibody responses induced by pathogen exposure. We identify predictive relationships between biomarkers and outcomes such as survival and pathogen shedding, and use these to impute values for missing data, thus increasing the size of the useable dataset. Mapping the host-pathogen space using quantitative biomarker data enables more nuanced understanding of an individual's time course of infection, duration of immunity, and probability of being infectious. Such maps also make efficient use of limited data for rare or poorly understood diseases, by providing a means to rapidly assess the range and extent of potential clinical and immunological profiles. These approaches yield benefits for clinicians needing to triage patients, prevent transmission, and assess immunity, and for disease ecologists or epidemiologists working to develop appropriate risk management strategies to reduce transmission risk on a population scale (e.g. model parameterization using more accurate estimates of duration of immunity and infectiousness) and to assess health impacts on a population scale.

Estimating prevalence from dried blood spots without using biological cut-offs: application of a novel approach to hepatitis C virus in drug users in France (ANRS-Coquelicot survey)

Seroprevalence estimation using cross-sectional serosurveys can be challenging due to inadequate or unknown biological cut-off limits of detection. In recent years, diagnostic assay cut-offs, fixed assay cut-offs and more flexible approaches as mixture modelling have been proposed to classify biological quantitative measurements into a positive or negative status. Our objective was to estimate the prevalence of anti-HCV antibodies among drug users (DU) in France in 2011 using a biological test performed on dried blood spots (DBS) collected during a cross-sectional serosurvey. However, in 2011, we did not have a cut-off value for DBS. We could not use the values for serum or plasma, knowing that the DBS value was not necessarily the same. Accordingly, we used a method which consisted of applying a two-component mixture model with age-dependent mixing proportions using penalised splines. The component densities were assumed to be log-normally distributed and were estimated in a Bayesian framework. Anti-HCV prevalence among DU was estimated at 43.3% in France and increased with age. Our method allowed us to provide estimates of age-dependent prevalence using DBS without having a specified biological cut-off value.

The seroprevalence of cytomegalovirus infection in Belgium anno 2002 and 2006: a comparative analysis with hepatitis A virus seroprevalence

Cytomegalovirus (CMV) infection is endemic worldwide but its seroprevalence varies widely. The goal of this study was to estimate the age-specific seroprevalence of CMV infection in Belgium based on two cross-sectional serological datasets from 2002 and 2006. The seroprevalence was estimated relying on diagnostic test results based on cut-off values pre-specified by the manufacturers of the tests as well as relying on mixture models applied to continuous pathogen-specific immunoglobulin G antibody titre concentrations. The age-specific seroprevalence of hepatitis A virus (HAV), based on three Belgian cross-sectional serological datasets from 1993, 2002 and 2006, was used as a comparator since individuals acquire lifelong immunity upon recovery, implying an increasing seroprevalence with age. The age group weighted overall CMV seroprevalence derived from the mixture model was 32% (95% confidence interval (CI) 31-34%) in 2002 and 31% (95% CI 30-32%) in 2006. We demonstrated that CMV epidemiology differs from the immunizing infection HAV. This was the first large-scale study of CMV and HAV serial datasets in Belgium, estimating seroprevalence specified by age and birth cohort.

Resurgence risk for measles, mumps and rubella in France in 2018 and 2020

Large measles and mumps outbreaks recently occurred throughout Europe and the United States. Aim: Our aim was to estimate and map the risk of resurgence for measles, mumps and rubella in France. Methods: We used a multi-cohort model combining seroprevalence information, vaccine coverage and social contact data. Results: The overall outbreak risk for France in 2018 was highest for mumps, remained significant for measles despite a recent measles outbreak and was low for rubella. Outbreak risks were heterogeneous between departments, as the effective reproduction numbers for 2018 ranged from 1.08 to 3.66. The seroprevalence, and therefore the risk of measles and rubella infection, differed significantly between males and females. There was a lower seroprevalence, and therefore a higher risk, for males. Infants of less than 1 year would be seriously affected in a future outbreak of measles, mumps or rubella, but the highest overall caseload contribution would come from teenagers and young adults (10–25 years old). Conclusions: The high risk for teenagers and young adults is of concern in view of their vulnerability to more severe measles, mumps and rubella disease and complications.

The Certainty of Uncertainty: Potential Sources of Bias and Imprecision in Disease Ecology Studies

Wildlife diseases have important implications for wildlife and human health, the preservation of biodiversity and the resilience of ecosystems. However, understanding disease dynamics and the impacts of pathogens in wild populations is challenging because these complex systems can rarely, if ever, be observed without error. Uncertainty in disease ecology studies is commonly defined in terms of either heterogeneity in detectability (due to variation in the probability of encountering, capturing, or detecting individuals in their natural habitat) or uncertainty in disease state assignment (due to misclassification errors or incomplete information). In reality, however, uncertainty in disease ecology studies extends beyond these components of observation error and can arise from multiple varied processes, each of which can lead to bias and a lack of precision in parameter estimates. Here, we present an inventory of the sources of potential uncertainty in studies that attempt to quantify disease-relevant parameters from wild populations (e.g., prevalence, incidence, transmission rates, force of infection, risk of infection, persistence times, and disease-induced impacts). We show that uncertainty can arise via processes pertaining to aspects of the disease system, the study design, the methods used to study the system, and the state of knowledge of the system, and that uncertainties generated via one process can propagate through to others because of interactions between the numerous biological, methodological and environmental factors at play. We show that many of these sources of uncertainty may not be immediately apparent to researchers (for example, unidentified crypticity among vectors, hosts or pathogens, a mismatch between the temporal scale of sampling and disease dynamics, demographic or social misclassification), and thus have received comparatively little consideration in the literature to date. Finally, we discuss the type of bias or imprecision introduced by these varied sources of uncertainty and briefly present appropriate sampling and analytical methods to account for, or minimise, their influence on estimates of disease-relevant parameters. This review should assist researchers and practitioners to navigate the pitfalls of uncertainty in wildlife disease ecology studies.

Inferring infection hazard in wildlife populations by linking data across individual and population scales

Our ability to infer unobservable disease-dynamic processes such as force of infection (infection hazard for susceptible hosts) has transformed our understanding of disease transmission mechanisms and capacity to predict disease dynamics. Conventional methods for inferring FOI estimate a time-averaged value and are based on population-level processes. Because many pathogens exhibit epidemic cycling and FOI is the result of processes acting across the scales of individuals and populations, a flexible framework that extends to epidemic dynamics and links within-host processes to FOI is needed. Specifically, within-host antibody kinetics in wildlife hosts can be short-lived and produce patterns that are repeatable across individuals, suggesting individual-level antibody concentrations could be used to infer time since infection and hence FOI. Using simulations and case studies (influenza A in lesser snow geese and Yersinia pestis in coyotes), we argue that with careful experimental and surveillance design, the population-level FOI signal can be recovered from individual-level antibody kinetics, despite substantial individual-level variation. In addition to improving inference, the cross-scale quantitative antibody approach we describe can reveal insights into drivers of individual-based variation in disease response, and the role of poorly understood processes such as secondary infections, in population-level dynamics of disease.

Age- and time-dependent prevalence and incidence of hepatitis C virus infection in drug users in France, 2004-2011: model-based estimation from two national cross-sectional serosurveys

Hepatitis C virus (HCV) infection is a public health issue worldwide. Injecting drug use remains the major mode of transmission in developed countries. Monitoring the HCV transmission dynamic over time is crucial, especially to assess the effect of harm reduction measures in drug users (DU). Our objective was to estimate the prevalence and incidence of HCV infection in DU in France using data from a repeated cross-sectional survey conducted in 2004 and 2011. Age- and time-dependent HCV prevalence was estimated through logistic regression models adjusted for HIV serostatus or injecting practices. HCV incidence was estimated from a mathematical model linking prevalence and incidence. HCV prevalence decreased from 58·2% [95% confidence interval (CI) 49·7-66·8] in 2004 to 43·2% (95% CI 38·8-47·7) in 2011. HCV incidence decreased from 7·9/100 person-years (95% CI 6·4-9·4) in 2004 to 4·4/100 person-years (95% CI 3·3-5·9) in 2011. HCV prevalence and incidence were significantly associated with age, calendar time, HIV serostatus and injecting practices. In 2011, the highest estimated incidence was in active injecting DU (11·2/100 person-years). Given the forthcoming objective of generalizing access to new direct antiviral agents for HCV infection, our results contribute to decision-making and policy development regarding treatment scale-up and disease prevention in the DU population.

Estimating Age-Specific Immunity and Force of Infection of Varicella Zoster Virus in Norway Using Mixture Models

This study applies mixture modelling to examine age-specific immunity to varicella zoster virus (VZV) infection in Norway based on the first large-scale serological study in the general population. We estimated the seropositive proportions at different ages and calculated the underlying force of infection by using a sample of 2103 residual sera obtained from patients seeking primary and hospital care. A rapid increase in the VZV-associated immunity is observed in the first years of life with 63% of children being immune by age 5. The increase in the immunity levels slows down thereafter, with a large proportion of adults still susceptible by age 20 (around 14.5%), thus at risk of serious sequelae of varicella infection. The corresponding force of infection peaks during the preschool period, subsequently declines to a minimum between ages 10 and 20 years, and afterwards moderately increases to reach a plateau lasting throughout the childbearing period. In comparison with the traditional cut-off approach, mixture modelling used the whole data without producing any inconclusive cases, led to an unbiased classification of individuals between susceptible and immune, and provided a smoother immune profile by age. These findings represent an important step towards any decision about the introduction of varicella vaccination in Norway, as they are a primary input for mathematical transmission models aimed at evaluating potential vaccination scenarios.

Estimating Time of Infection Using Prior Serological and Individual Information Can Greatly Improve Incidence Estimation of Human and Wildlife Infections

Diseases of humans and wildlife are typically tracked and studied through incidence, the number of new infections per time unit. Estimating incidence is not without difficulties, as asymptomatic infections, low sampling intervals and low sample sizes can introduce large estimation errors. After infection, biomarkers such as antibodies or pathogens often change predictably over time, and this temporal pattern can contain information about the time since infection that could improve incidence estimation. Antibody level and avidity have been used to estimate time since infection and to recreate incidence, but the errors on these estimates using currently existing methods are generally large. Using a semi-parametric model in a Bayesian framework, we introduce a method that allows the use of multiple sources of information (such as antibody level, pathogen presence in different organs, individual age, season) for estimating individual time since infection. When sufficient background data are available, this method can greatly improve incidence estimation, which we show using arenavirus infection in multimammate mice as a test case. The method performs well, especially compared to the situation in which seroconversion events between sampling sessions are the main data source. The possibility to implement several sources of information allows the use of data that are in many cases already available, which means that existing incidence data can be improved without the need for additional sampling efforts or laboratory assays.

Estimating malaria transmission intensity from Plasmodium falciparum serological data using antibody density models

Background

Serological data are increasingly being used to monitor malaria transmission intensity and have been demonstrated to be particularly useful in areas of low transmission where traditional measures such as EIR and parasite prevalence are limited. The seroconversion rate (SCR) is usually estimated using catalytic models in which the measured antibody levels are used to categorize individuals as seropositive or seronegative. One limitation of this approach is the requirement to impose a fixed cut-off to distinguish seropositive and negative individuals. Furthermore, the continuous variation in antibody levels is ignored thereby potentially reducing the precision of the estimate.

Methods

An age-specific density model which mimics antibody acquisition and loss was developed to make full use of the information provided by serological measures of antibody levels. This was fitted to blood-stage antibody density data from 12 villages at varying transmission intensity in Northern Tanzania to estimate the exposure rate as an alternative measure of transmission intensity.

Results

The results show a high correlation between the exposure rate estimates obtained and the estimated SCR obtained from a catalytic model (r = 0.95) and with two derived measures of EIR (r = 0.74 and r = 0.81). Estimates of exposure rate obtained with the density model were also more precise than those derived from catalytic models.

Conclusion

This approach, if validated across different epidemiological settings, could be a useful alternative framework for quantifying transmission intensity, which makes more complete use of serological data.

Assessing mumps outbreak risk in highly vaccinated populations using spatial seroprevalence data

Mumps is a potentially severe viral infection. The incidence of mumps has declined dramatically in high-income countries since the introduction of mumps antigen-containing vaccines. However, recent large outbreaks of mumps in highly vaccinated populations suggest waning of vaccine-induced immunity and primary vaccine failure. In this paper we present a simple method for identifying geographic regions with high outbreak potential, demonstrated using 2006 mumps seroprevalence data from Belgium and Belgian vaccination coverage data. Predictions of the outbreak potential in terms of the effective reproduction number in future years signal an increased risk of new mumps outbreaks. Literature reviews on serological information for both primary vaccine failure and waning immunity provide essential information for our predictions. Tailor-made additional vaccination campaigns would be valuable for decreasing local pockets of susceptibility, thereby reducing the risk of future large-scale mumps outbreaks.