Geographical variation of COVID-19 vaccination coverage, ethnic diversity and population composition in Flanders.

The vaccination coverage in Flanders is high, but some regions show lower vaccination willingness as compared to the overall vaccination coverage. Beginning November of 2021, the vaccination rate in Flanders was above 93% in age groups above 45 years, and around 85% in the age groups 12 to 44 years. Apart from Flanders as a whole, focus here is on the health sector Maasland, which has a slightly lower vaccination rate, especially in the age groups 12 to 44 years. In the Maasland region, located on the eastern border of Flanders, there are between 1% and 10% less vaccinated individuals than expected according to the vaccination rate in the whole of Flanders, with lowest vaccination rates in the south of the Maasland region. We study the impact of ethnic diversity in the population, population composition with respect to the ethnicity of individuals (in the sense of how the local population composition differs from the Flemish average), and socio-economic status on the vaccination rate at the level of the statistical sector, apart from the effect of age. We explain the statistical methods to investigate geographical differences and illustrate how one can deal with incomplete information in vaccination registries. Ethnic diversity in a region is associated with lower vaccination rates, as is a lower regional socio-economic status. The composition of the population in Maasland is associated with a 35% reduction in the odds to get vaccinated as compared to the overall Flemish population.

Comparison of different software implementations for spatial disease mapping.

Disease mapping is a scientific field that aims to understand and predict disease risk based on counts of observed cases within small regions of a study area of interest. Hierarchical model-based approaches that borrow information from neighbouring areas via conditional autoregressive (CAR) random effects on the local disease rates have gained a lot of popularity, thanks to the readily implemented Markov chain Monte Carlo methods. Nowadays, many software implementations to model risk distributions exist. Many of these applications differ, to varying degrees, in the underlying methodology. This paper provides an in-depth comparison between analysis results, coming from R-packages CARBayes, R2OpenBUGS, NIMBLE, R2BayesX, R-INLA, and RStan. We investigate CAR models typically used in disease mapping for spatially discrete count data. Data about diabetics in children and young adults in Belgium are used in a case study, while simulation studies are undertaken to assess software performance in different settings.

Spatial smoothing models to deal with the complex sampling design and nonresponse in the Florida BRFSS survey.

Public health and governmental organizations have acknowledged the importance of obtaining information of various characteristics for small areas, such as counties. Spatial smoothing models have been developed to gain reliable information on the geographical distribution of the outcome of interest. When the geographical analysis is based on survey data, two issues pose challenges: (1) the complex design of the survey and (2) the presence of missing data due to non-response. We investigate the influence of missing data and the adjustment thereof in the context of the 2013 Florida Behavioral Risk Factor Surveillance System (BRFSS) health survey. We focus on the application and comparison of the Hajek ratio estimator and two model-based approaches for estimation of the spatial trend of the prevalence of having no health insurance coverage. The model-based methods are compared using the Deviance Information Criterion which show the benefits of modeling the weights as flexibly as possible. Methods are extended towards subgroup analyses and the estimation of area-specific standardized rates, where household incomes was identified as an important factor to include in the analysis.

Spatiotemporal multivariate mixture models for Bayesian model selection in disease mapping.

It is often the case that researchers wish to simultaneously explore the behavior of and estimate overall risk for multiple, related diseases with varying rarity while accounting for potential spatial and/or temporal correlation. In this paper, we propose a flexible class of multivariate spatio-temporal mixture models to fill this role. Further, these models offer flexibility with the potential for model selection as well as the ability to accommodate lifestyle, socio-economic, and physical environmental variables with spatial, temporal, or both structures. Here, we explore the capability of this approach via a large scale simulation study and examine a motivating data example involving three cancers in South Carolina. The results which are focused on four model variants suggest that all models possess the ability to recover simulation ground truth and display improved model fit over two baseline Knorr-Held spatio-temporal interaction model variants in a real data application.

Spatial small area smoothing models for handling survey data with nonresponse.

Spatial smoothing models play an important role in the field of small area estimation. In the context of complex survey designs, the use of design weights is indispensable in the estimation process. Recently, efforts have been made in these spatial smoothing models, in order to obtain reliable estimates of the spatial trend. However, the concept of missing data remains a prevalent problem in the context of spatial trend estimation as estimates are potentially subject to bias. In this paper, we focus on spatial health surveys where the available information consists of a binary response and its associated design weight. Furthermore, we investigate the impact of nonresponse as missing data on a range of spatial models for different missingness mechanisms and different degrees of missingness by means of an extensive simulation study. The computations were performed in R, using INLA and other existing packages. The results show that weight adjustment to correct for missingness has a beneficial effect on the bias in the missing at random setting for all models. Furthermore, we estimate the geographical distribution of perceived health at the district level based on the Belgian Health Interview Survey (2001). Copyright © 2017 John Wiley & Sons, Ltd.

Spatio-temporal Bayesian model selection for disease mapping.

Spatio-temporal analysis of small area health data often involves choosing a fixed set of predictors prior to the final model fit. In this paper, we propose a spatio-temporal approach of Bayesian model selection to implement model selection for certain areas of the study region as well as certain years in the study time line. Here, we examine the usefulness of this approach by way of a large-scale simulation study accompanied by a case study. Our results suggest that a special case of the model selection methods, a mixture model allowing a weight parameter to indicate if the appropriate linear predictor is spatial, spatio-temporal, or a mixture of the two, offers the best option to fitting these spatio-temporal models. In addition, the case study illustrates the effectiveness of this mixture model within the model selection setting by easily accommodating lifestyle, socio-economic, and physical environmental variables to select a predominantly spatio-temporal linear predictor.

Model-based inference for small area estimation with sampling weights.

Obtaining reliable estimates about health outcomes for areas or domains where only few to no samples are available is the goal of small area estimation (SAE). Often, we rely on health surveys to obtain information about health outcomes. Such surveys are often characterised by a complex design, stratification, and unequal sampling weights as common features. Hierarchical Bayesian models are well recognised in SAE as a spatial smoothing method, but often ignore the sampling weights that reflect the complex sampling design. In this paper, we focus on data obtained from a health survey where the sampling weights of the sampled individuals are the only information available about the design. We develop a predictive model-based approach to estimate the prevalence of a binary outcome for both the sampled and non-sampled individuals, using hierarchical Bayesian models that take into account the sampling weights. A simulation study is carried out to compare the performance of our proposed method with other established methods. The results indicate that our proposed method achieves great reductions in mean squared error when compared with standard approaches. It performs equally well or better when compared with more elaborate methods when there is a relationship between the responses and the sampling weights. The proposed method is applied to estimate asthma prevalence across districts.

Comparing INLA and OpenBUGS for hierarchical Poisson modeling in disease mapping.

The recently developed R package INLA (Integrated Nested Laplace Approximation) is becoming a more widely used package for Bayesian inference. The INLA software has been promoted as a fast alternative to MCMC for disease mapping applications. Here, we compare the INLA package to the MCMC approach by way of the BRugs package in R, which calls OpenBUGS. We focus on the Poisson data model commonly used for disease mapping. Ultimately, INLA is a computationally efficient way of implementing Bayesian methods and returns nearly identical estimates for fixed parameters in comparison to OpenBUGS, but falls short in recovering the true estimates for the random effects, their precisions, and model goodness of fit measures under the default settings. We assumed default settings for ground truth parameters, and through altering these default settings in our simulation study, we were able to recover estimates comparable to those produced in OpenBUGS under the same assumptions.

The social contact hypothesis under the assumption of endemic equilibrium: Elucidating the transmission potential of VZV in Europe.

The basic reproduction number R0 and the effective reproduction number R are pivotal parameters in infectious disease epidemiology, quantifying the transmission potential of an infection in a population. We estimate both parameters from 13 pre-vaccination serological data sets on varicella zoster virus (VZV) in 12 European countries and from population-based social contact surveys under the commonly made assumptions of endemic and demographic equilibrium. The fit to the serology is evaluated using the inferred effective reproduction number R as a model eligibility criterion combined with AIC as a model selection criterion. For only 2 out of 12 countries, the common choice of a constant proportionality factor is sufficient to provide a good fit to the seroprevalence data. For the other countries, an age-specific proportionality factor provides a better fit, assuming physical contacts lasting longer than 15 min are a good proxy for potential varicella transmission events. In all countries, primary infection with VZV most often occurs in early childhood, but there is substantial variation in transmission potential with R0 ranging from 2.8 in England and Wales to 7.6 in The Netherlands. Two non-parametric methods, the maximal information coefficient (MIC) and a random forest approach, are used to explain these differences in R0 in terms of relevant country-specific characteristics. Our results suggest an association with three general factors: inequality in wealth, infant vaccination coverage and child care attendance. This illustrates the need to consider fundamental differences between European countries when formulating and parameterizing infectious disease models.

Serological diagnosis of bovine neosporosis: a Bayesian evaluation of two antibody ELISA tests for in vivo diagnosis in purchased and abortion cattle.

Currently, there are no perfect reference tests for the in vivo detection of Neospora caninum infection. Two commercial N caninum ELISA tests are currently used in Belgium for bovine sera (TEST A and TEST B). The goal of this study is to evaluate these tests used at their current cut-offs, with a no gold standard approach, for the test purpose of (1) demonstration of freedom of infection at purchase and (2) diagnosis in aborting cattle. Sera of two study populations, Abortion population (n=196) and Purchase population (n=514), were selected and tested with both ELISA's. Test results were entered in a Bayesian model with informative priors on population prevalences only (Scenario 1). As sensitivity analysis, two more models were used: one with informative priors on test diagnostic accuracy (Scenario 2) and one with all priors uninformative (Scenario 3). The accuracy parameters were estimated from the first model: diagnostic sensitivity (Test A: 93.54 per cent-Test B: 86.99 per cent) and specificity (Test A: 90.22 per cent-Test B: 90.15 per cent) were high and comparable (Bayesian P values >0.05). Based on predictive values in the two study populations, both tests were fit for purpose, despite an expected false negative fraction of ±0.5 per cent in the Purchase population and ±5 per cent in the Abortion population. In addition, a false positive fraction of ±3 per cent in the overall Purchase population and ±4 per cent in the overall Abortion population was found.

Seventy-five years of estimating the force of infection from current status data.

The force of infection, describing the rate at which a susceptible person acquires an infection, is a key parameter in models estimating the infectious disease burden, and the effectiveness and cost-effectiveness of infectious disease prevention. Since Muench formulated the first catalytic model to estimate the force of infection from current status data in 1934, exactly 75 years ago, several authors addressed the estimation of this parameter by more advanced statistical methods, while applying these to seroprevalence and reported incidence/case notification data. In this paper we present an historical overview, discussing the relevance of Muench's work, and we explain the wide array of newer methods with illustrations on pre-vaccination serological survey data of two airborne infections: rubella and parvovirus B19. We also provide guidance on deciding which method(s) to apply to estimate the force of infection, given a particular set of data.

Spatial analysis of breast and cervical cancer incidence in small geographical areas in Cuba, 1999-2003

According to the data from the National Cancer Registry, breast and cervical cancer are the two most common nonskin cancers in Cuban woman. This study was addressed to describe the geographical variation of their incidence at small area level over the period 1999-2003. For each municipality, standardized incidence ratios were calculated and smoothed using a Poisson-Gamma, Poisson-Lognormal and a Conditional Autoregressive (CAR) model. The covariate 'urbanization level' was included in the Poisson-Lognormal and CAR models. The posterior probability of each municipality's relative risk (RR) exceeding unity was computed. Clusters were confirmed using the spatial scan statistic of Kulldorff. The CAR model provided the best fit for the geographical distribution of breast and cervical cancer in Cuba. For breast cancer, a high-risk region was identified in municipalities of Ciudad de La Habana province (CAR-smoothed RR between 1.21 and 1.26). Cervical cancer exhibited two areas with excess risk in the east and extreme west of the island (CAR-smoothed RR range 1.2-2.01 both areas together). Clusters were confirmed only for cervical cancer (P = 0.001 for the most likely cluster and P = 0.003 for a secondary cluster). In conclusion, the study supports the hypothesis of a spatial variation in risk at small area level essentially for cervical cancer and also for breast cancer that probably reflects the territorial distribution of life style and socioeconomic factors. This is the first attempt to introduce this methodology in the framework of the National Cancer Registry of Cuba and we expect to extend its use to forthcoming analyses(AU)

Miscoding: a threat to the hospital care system. How to detect it?

BACKGROUND: Artificially influencing the case mix of hospitals may have several deleterious consequences for the hospital care system. One distinguishes over-evaluation (up-coding) and under-evaluation (under-coding) of the case mix. Apart from its financial consequences, miscoding may cause a fracture in epidemiological time series and, by increasing artificially the severity of illness, may affect the assessment of the quality of hospital care, based on administrative data. METHODS: Fixed effects models were used to assess deviant coding behavior at the hospital level. To do so, we examined the linear evolution over time of characteristics such as length of stay and of 21 "triggering" conditions susceptible to increase the case mix of a stay. In case of deviant coding, these triggering conditions were checked to direct the audit towards fraud-suspected discharge abstracts. Hereto, a method consisting in comparing a single hospital's linear evolution over time with the national linear evolution over time was developed, using an interaction term between linear evolution over time and hospitals. To test this methodology, fraud-directed audits were carried out in addition to the usual, at random audits. RESULTS: Important inter-hospital differences in the linear evolution over time of several characteristics of Belgian hospitals were identified, as well as evidence not only of improving coding practices, but also of up-coding, fraudulent under-coding and of numerous coding errors without financial impact. The coding errors, ascertained in the at random audit, resulted in a wrongful gain for the faulty hospitals of 28.23 days in 258 stays, whereas in case of fraud-directed audits these figures amounted up to 642.68 days in 334 stays. CONCLUSION: Fraud-directed audit may constitute a valuable tool in the quality assurance of administrative databases, improving their use in epidemiology and assessment of the quality of care.

Impact of human interventions on the spread of bluetongue virus serotype 8 during the 2006 epidemic in north-western Europe.

Bluetongue virus (BTV) can be spread by movement or migration of infected ruminants. Infected midges (Culicoides sp.) can be dispersed with livestock or on the wind. Transmissions of infection from host to host by semen and trans-placental infection of the embryo from the dam have been found. As for any infectious animal disease, the spread of BTV can be heavily influenced by human interventions preventing or facilitating the transmission pathways. This paper describes the results of investigations that were conducted on the potential role of the above-mentioned human interventions on the spread of BTV-8 during the 2006 epidemic in north-western Europe. Data on surveillance and control measures implemented in the affected European Union (EU) Member States (MS) were extracted from the legislation and procedures adopted by the national authorities in Belgium, France, Germany, and The Netherlands. The impact of the control measures on the BTV-incidence in time and space was explored. Data on ruminant transports leaving the area of first infection (AFI) to other areas within and beyond the affected MS were obtained from the national identification and registration systems of the three initially affected MS (Belgium, Germany, The Netherlands) and from the Trade Control and Expert System (TRACES) of the European Commission. The association between the cumulative number of cases that occurred in a municipality outside the AFI and the number of movements or the number of animals moved from the AFI to that municipality was assessed using a linear negative binomial regression model. The results of this study indicated that the control measures which were implemented in the affected MS (in accordance with EU directives) were not able to fully stop further spread of BTV and to control the epidemic. This finding is not surprising because BT is a vector-borne disease and it is difficult to limit vector movements. We could not assess the consequences of not taking control measures at all but it is possible, if not most likely, that this would have resulted in even wider spread. The study also showed an indication of the possible involvement of animal movements in the spread of BTV during the epidemic. Therefore, the prevention of animal movements remains an important tool to control BTV outbreaks. The extension of the epidemic to the east cannot be explained by the movement of animals, which mainly occurred in a north-western direction. This indicates that it is important to consider other influential factors such as dispersal of infected vectors depending on wind direction, or local spread.

Establishing the spread of bluetongue virus at the end of the 2006 epidemic in Belgium.

Bluetongue (BT) was notified for the first time in several Northern European countries in August 2006. The first reported outbreaks of BT were confirmed in herds located near the place where Belgium, The Netherlands and Germany share borders. The disease was rapidly and widely disseminated throughout Belgium in both sheep and cattle herds. During the epidemic, case reporting by the Veterinary Authorities relied almost exclusively on the identification of herds with confirmed clinical infected ruminants. A cross-sectional serological survey targeting all Belgian ruminants was then undertaken during the vector-free season. The first objective of this study was to provide unbiased estimates of BT-seroprevalence for different regions of Belgium. Since under-reporting was suspected during the epidemic, a second goal was to compare the final dispersion of the virus based on the seroprevalence estimates to the dispersion of the confirmed clinical cases which were notified in Belgium, in order to estimate the accuracy of the case detection based on clinical suspicion. True within-herd seroprevalence was estimated based on a logistic-normal regression model with prior specification on the diagnostic test's sensitivity and specificity. The model was fitted in a Bayesian framework. Herd seroprevalence was estimated using a logistic regression model. To study the linear correlation between the BT winter screening data and the case-herds data, the linear predicted values for the herd prevalence were compared and the Pearson correlation coefficient was estimated. The overall herd and true within-herd seroprevalences were estimated at 83.3 (79.2-87.0) and 23.8 (20.1-28.1)%, respectively. BT seropositivity was shown to be widely but unevenly distributed throughout Belgium, with a gradient decreasing towards the south and the west of the country. The analysis has shown there was a strong correlation between the outbreak data and the data from the survey (r=0.73, p<0.0001). The case detection system based on clinical suspicion underestimated the real impact of the epidemic, but indicated an accurate spatial distribution of the virus at the end of the epidemic.

On the use of historical control data in pre-clinical safety studies.

A number of methods to formally incorporate historical control information in pre-clinical safety evaluation studies have been proposed in literature. However, it remains unclear when one should use historical data. Focusing on the logistic-normal model, we investigate situations where historical studies may prove to be useful. Aspects of estimation (precision and bias) and testing (power) for treatment effect are investigated under different conditions such as the number of historical control studies, the degree of homogeneity amongst them, the level of treatment effect and different control rates. The possibility to use a selected subset of historical control studies is also explored.

Spatial determination of successive spikes in the isolated cat duodenum.

In seven isolated segments of the feline duodenum, the timings of all spikes and the locations of all spike patches that occurred after 12-16 successive slow waves were analysed. Simultaneous recordings were performed during 1-min periods using 240 extracellular electrodes (24 x 10 array; interelectrode distance 2 mm) positioned onto the serosal surface. In all seven preparations, spikes always occurred during the first half of the slow wave cycle. From preparation to preparation, and within 1-min periods in each preparation, there was limited variation in the spike-spike intervals, in the times between the spikes and the preceding slow wave and in the number of spikes at each electrode site. In contrast, the number of electrode sites that recorded spikes and the number of spike patches both showed great variability between preparations and sometimes within a single preparation. In addition, the location of spikes and spike patches was not random but was significantly concentrated in certain areas, often located along the anti-mesenteric border, while other sites showed little or no spike activity. In conclusion, spikes and spike patches tend to occur significantly in some areas and not in others. This spatial heterogeneity will play a role in intestinal motility.