Temporal, spatial and space-time distribution of infections caused by five major enteric pathogens, Ontario, Canada, 2010-2017

AIMS

In Canada, enteric diseases pose substantial health and economic burdens. The distribution of these diseases is uneven across both geography and time and understanding these patterns is therefore important for the prevention of future outbreaks. We evaluated temporal, spatial and space-time clustering of laboratory-confirmed cases of Campylobacter spp. (n = 28,728), non-typhoidal Salmonella spp. (n = 22,640), Shiga toxin-producing Escherichia coli (STEC; n = 1340), Yersinia spp. (n = 1674) and Listeria monocytogenes (n = 471) infections, reported between 2010 and 2017 inclusive in Ontario, the most populous province in Canada (population ~ 13,500,000 in 2016).

METHODS AND RESULTS

For each enteric pathogen, we calculated the mean incidence rates (IRs) for Ontario's 35 public health unit (PHU) areas and visualized them using choropleth maps. We identified temporal, spatial and space-time high infection rate clusters using retrospective Poisson scan statistics. Campylobacter and Salmonella infections had the highest IRs, while Listeria infections had the lowest. Campylobacter, Salmonella, STEC and Listeria mostly clustered temporally in the spring/summer and sometimes extended into fall, while Yersinia showed a less clear seasonal pattern. The IR visualizations and spatial and space-time scan statistics showed geographic heterogeneity of infection rates with high infection rate clusters detected mainly in PHUs across the southwestern and central-western regions of Ontario for Campylobacter, Salmonella and STEC infections, and mainly in PHUs located in the central-eastern regions for Yersinia and Listeria. A high proportion of cases in some of the significant Salmonella, STEC and Listeria infection clusters were linked to disease outbreaks.

CONCLUSIONS

Results from this study will inform heightened public health surveillance, and prevention and control programmes, in populations and regions of high infection rates. Further research is needed to determine the pathogen-specific socioeconomic, environmental and agricultural risk factors that may be related to the temporal and spatial disease patterns we observed in our study.

An open-source implementation of tree-based scan statistics

PURPOSE

We develop an open-source R package to implement tree-based scan statistics (TBSS) analyses.

METHODS

TBSS are data mining methods used by the United States Food and Drug Administration and the Centers for Disease Control. They simultaneously screen thousands of hierarchically aggregated outcomes to identify unsuspected adverse effects of drugs or vaccines, accounting for multiple comparisons. The general structure of TBSS is highly adaptable, with four essential components: (1) a hierarchical outcome structure, (2) a test statistic to be computed for each element of the hierarchy, (3) an algorithm to generate data replicates under a null distribution, and (4) observed outcomes at the lower level of the hierarchy. We encode the general TBSS framework in a convenient R package that offers user-friendly functions for the most used TBSS methods. To illustrate the performance of our software, we evaluated two examples of archetypical TBSS analyses previously analyzed using proprietary, closed-source TreeScan™ software. The first considers the risk of congenital malformations associated with first-trimester exposure to valproate, and the second compares exposure to newly prescribed canagliflozin with a dipeptidyl peptidase 4 inhibitor in adults affected by type 2 diabetes.

RESULTS

The results of the original studies are replicated.

CONCLUSIONS

The diffusion of an open-source implementation of TBSS can enhance innovation of TBSS methods and foster collaborations. We offer an intuitive R package implementing standard TBSS methods with accompanying tutorials. Our unified object-oriented implementation allows expert users to extend the framework, introduce new features, or enhance existing ones.

Space-time self-harm and suicide clusters in two cities in Taiwan

AIMS

Suicidal acts may cluster in time and space and lead to community concerns about further imitative suicidal episodes. Although suicide clusters have been researched in previous studies, less is known about the clustering of non-fatal suicidal behaviour (self-harm). Furthermore, most previous studies used crude temporal and spatial information, e.g., numbers aggregated by month and residence area, for cluster detection analysis. This study aimed to (i) identify space-time clusters of self-harm and suicide using daily incidence data and exact address and (ii) investigate the characteristics of cluster-related suicidal acts.

METHODS

Data on emergency department presentations for self-harm and suicide deaths in Taipei City and New Taipei City, Taiwan, were used in this study. In all-age and age-specific analyses, self-harm and suicide clusters were identified using space-time permutation scan statistics. A cut-off of 0.10 for the p value was used to identify possible clusters. Logistic regression was used to investigate the characteristics associated with cluster-related episodes.

RESULTS

A total of 5,291 self-harm episodes and 1,406 suicides in Taipei City (2004-2006) and 20,531 self-harm episodes and 2,329 suicides in New Taipei City (2012-2016) were included in the analysis. In the two cities, two self-harm clusters (n [number of self-harm episodes or suicide deaths in the cluster] = 4 and 8 in Taipei City), four suicide clusters (n = 3 in Taipei City and n = 4, 11 and 4 in New Taipei City) and two self-harm and suicide combined clusters (n = 4 in Taipei City and n = 8 in New Taipei City) were identified. Space-time clusters of self-harm, suicide, and self-harm and suicide combined accounted for 0.05%, 0.59%, and 0.08% of the respective groups of suicidal acts. Cluster-related episodes of self-harm and suicide were more likely to be male (adjusted odds ratio [aOR] = 2.22, 95% confidence interval [CI] 1.26, 3.89) and young people aged 10-29 years (aOR = 2.72, 95% CI 1.43, 5.21) than their cluster-unrelated counterparts.

CONCLUSIONS

Space-time clusters of self-harm, suicide, and self-harm and suicide combined accounted for a relatively small proportion of suicidal acts and were associated with some sex/age characteristics. Focusing on suicide deaths alone may underestimate the size of some clusters and/or lead to some clusters being overlooked. Future research could consider combining self-harm and suicide data and use social connection information to investigate possible clusters of suicidal acts.

Identifying hidden Zika hotspots in Pernambuco, Brazil: a spatial analysis

BACKGROUND

Northeast Brazil has the world's highest rate of Zika-related microcephaly. However, Zika case counts cannot accurately describe burden because mandatory reporting was only established when the epidemic was declining in the region.

METHODS

To advance the study of the Zika epidemic, we identified hotspots of Zika in Pernambuco state, Northeast Brazil, using Aedes-borne diseases (dengue, chikungunya and Zika) and microcephaly data. We used Kulldorff's Poisson purely spatial scan statistic to detect low- and high-risk clusters for Aedes-borne diseases (2014-2017) and for microcephaly (2015-2017), separately. Municipalities were classified according to a proposed gradient of Zika burden during the epidemic, based on the combination of cluster status in each analysis and considering the strength of the evidence.

RESULTS

We identified 26 Aedes-borne diseases clusters (11 high-risk) and 5 microcephaly clusters (3 high-risk) in Pernambuco. According to the proposed Zika burden gradient, our results indicate that the northeast of Pernambuco and the Sertão region were hit hardest by the Zika epidemic. The first is the most populous area of Pernambuco, while the second has one of the highest rates of social and economic inequality in Brazil.

CONCLUSION

We successfully identified possible hidden Zika hotspots using a simple methodology combining Aedes-borne diseases and microcephaly information.

Identification of geographic clusters for temporal heterogeneity with application to dengue surveillance

Identifying transmission of hot spots with temporal trends is important for reducing infectious disease propagation. Cluster analysis is a particularly useful tool to explore underlying stochastic processes between observations by grouping items into categories by their similarity. In a study of epidemic propagation, clustering geographic regions that have similar time series could help researchers track diffusion routes from a common source of an infectious disease. In this article, we propose a two‐stage scan statistic to classify regions into various geographic clusters by their temporal heterogeneity. The proposed scan statistic is more flexible than traditional methods in that contiguous and nonproximate regions with similar temporal patterns can be identified simultaneously. A simulation study and data analysis for a dengue fever infection are also presented for illustration.

Geospatial dynamics of COVID-19 clusters and hotspots in Bangladesh

The coronavirus disease 2019 (COVID-19) is an emerging and rapidly evolving profound pandemic, which causes severe acute respiratory syndrome and results in significant case fatality around the world including Bangladesh. We conducted this study to assess how COVID-19 cases clustered across districts in Bangladesh and whether the pattern and duration of clusters changed following the country's containment strategy using Geographic information system (GIS) software. We calculated the epidemiological measures including incidence, case fatality rate (CFR) and spatiotemporal pattern of COVID-19. We used inverse distance weighting (IDW), Geographically weighted regression (GWR), Moran's I and Getis-Ord Gi* statistics for prediction, spatial autocorrelation and hotspot identification. We used retrospective space-time scan statistic to analyse clusters of COVID-19 cases. COVID-19 has a CFR of 1.4%. Over 50% of cases were reported among young adults (21-40 years age). The incidence varies from 0.03 - 0.95 at the end of March to 15.59-308.62 per 100,000, at the end of July. Global Moran's Index indicates a robust spatial autocorrelation of COVID-19 cases. Local Moran's I analysis stated a distinct High-High (HH) clustering of COVID-19 cases among Dhaka, Gazipur and Narayanganj districts. Twelve statistically significant high rated clusters were identified by space-time scan statistics using a discrete Poisson model. IDW predicted the cases at the undetermined area, and GWR showed a strong relationship between population density and case frequency, which was further established with Moran's I (0.734; p ≤ 0.01). Dhaka and its surrounding six districts were identified as the significant hotspot whereas Chattogram was an extended infected area, indicating the gradual spread of the virus to peripheral districts. This study provides novel insights into the geostatistical analysis of COVID-19 clusters and hotspots that might assist the policy planner to predict the spatiotemporal transmission dynamics and formulate imperative control strategies of SARS-CoV-2 in Bangladesh. The geospatial modeling tools can be used to prevent and control future epidemics and pandemics.

STRUCTURED CORRELATION DETECTION WITH APPLICATION TO COLOCALIZATION ANALYSIS IN DUAL-CHANNEL FLUORESCENCE MICROSCOPIC IMAGING

Current workflows for colocalization analysis in fluorescence microscopic imaging introduce significant bias in terms of the user's choice of region of interest (ROI). In this work, we introduce an automatic, unbiased structured detection method for correlated region detection between two random processes observed on a common domain. We argue that although intuitive, using the maximum log-likelihood statistic directly suffers from potential bias and substantially reduced power. Therefore, we introduce a simple size-based normalization to overcome this problem. We show that scanning using the proposed statistic leads to optimal correlated region detection over a large collection of structured correlation detection problems.

Dynamic Scan Procedure for Detecting Rare-Variant Association Regions in Whole-Genome Sequencing Studies

Whole-genome sequencing (WGS) studies are being widely conducted in order to identify rare variants associated with human diseases and disease-related traits. Classical single-marker association analyses for rare variants have limited power, and variant-set-based analyses are commonly used by researchers for analyzing rare variants. However, existing variant-set-based approaches need to pre-specify genetic regions for analysis; hence, they are not directly applicable to WGS data because of the large number of intergenic and intron regions that consist of a massive number of non-coding variants. The commonly used sliding-window method requires the pre-specification of fixed window sizes, which are often unknown as a priori, are difficult to specify in practice, and are subject to limitations given that the sizes of genetic-association regions are likely to vary across the genome and phenotypes. We propose a computationally efficient and dynamic scan-statistic method (Scan the Genome [SCANG]) for analyzing WGS data; this method flexibly detects the sizes and the locations of rare-variant association regions without the need to specify a prior, fixed window size. The proposed method controls for the genome-wise type I error rate and accounts for the linkage disequilibrium among genetic variants. It allows the detected sizes of rare-variant association regions to vary across the genome. Through extensive simulated studies that consider a wide variety of scenarios, we show that SCANG substantially outperforms several alternative methods for detecting rare-variant-associations while controlling for the genome-wise type I error rates. We illustrate SCANG by analyzing the WGS lipids data from the Atherosclerosis Risk in Communities (ARIC) study.

The Use of Spatial and Spatiotemporal Modeling for Surveillance of H5N1 Highly Pathogenic Avian Influenza in Poultry in the Middle East

Since 2005, H5N1 highly pathogenic avian influenza virus (HPAIV) has severely impacted the economy and public health in the Middle East (ME) with Egypt as the most affected country. Understanding the high-risk areas and spatiotemporal distribution of the H5N1 HPAIV in poultry is prerequisite for establishing risk-based surveillance activities at a regional level in the ME. Here, we aimed to predict the geographic range of H5N1 HPAIV outbreaks in poultry in the ME using a set of environmental variables and to investigate the spatiotemporal clustering of outbreaks in the region. Data from the ME for the period 2005-14 were analyzed using maximum entropy ecological niche modeling and the permutation model of the scan statistics. The predicted range of high-risk areas (P > 0.60) for H5N1 HPAIV in poultry included parts of the ME northeastern countries, whereas the Egyptian Nile delta and valley were estimated to be the most suitable locations for occurrence of H5N1 HPAIV outbreaks. The most important environmental predictor that contributed to risk for H5N1 HPAIV was the precipitation of the warmest quarter (47.2%), followed by the type of global livestock production system (18.1%). Most significant spatiotemporal clusters (P < 0.001) were detected in Egypt, Turkey, Kuwait, Saudi Arabia, and Sudan. Results suggest that more information related to poultry holding demographics is needed to further improve prediction of risk for H5N1 HPAIV in the ME, whereas the methodology presented here may be useful in guiding the design of surveillance programs and in identifying areas in which underreporting may have occurred.

Potentially inappropriate medication prescribing is associated with socioeconomic factors: a spatial analysis in the French Nord-Pas-de-Calais Region

Background

potentially inappropriate medication (PIM) prescribing is common in older people and leads to adverse events and hospital admissions.

Objective

to determine whether prevalence of PIM prescribing varies according to healthcare supply and socioeconomic status.

Methods

all prescriptions dispensed at community pharmacies for patients aged 75 and older between 1 January  and 31 March 2012 were retrieved from French Health Insurance Information System of the Nord-Pas-de-Calais Region for patients affiliated to the Social Security scheme. PIM was defined according to the French list of Laroche. The geographic distribution of PIM prescribing in this area was analysed using spatial scan statistics.

Results

overall, 65.6% (n = 207,979) of people aged 75 years and over living in the Nord-Pas-de-Calais Region were included. Among them, 32.6% (n = 67,863) received at least one PIM. The spatial analysis identified 16 and 10 clusters of municipalities with a high and a low prevalence of PIM prescribing, respectively. Municipalities with a low prevalence of PIM were characterised by a high socioeconomic status whereas those with a high prevalence of PIM were mainly characterised by a low socioeconomic status, such as a high unemployment rate and low household incomes. Markers of healthcare supply were weakly associated with high or low prevalence clusters.

Conclusion

significant geographic variation in PIM prescribing was observed in the study territory and was mainly associated with socioeconomic factors.

Spatio-Temporal Distribution Characteristics and Trajectory Similarity Analysis of Tuberculosis in Beijing, China

Tuberculosis (TB) is an infectious disease with one of the highest reported incidences in China. The detection of the spatio-temporal distribution characteristics of TB is indicative of its prevention and control conditions. Trajectory similarity analysis detects variations and loopholes in prevention and provides urban public health officials and related decision makers more information for the allocation of public health resources and the formulation of prioritized health-related policies. This study analysed the spatio-temporal distribution characteristics of TB from 2009 to 2014 by utilizing spatial statistics, spatial autocorrelation analysis, and space-time scan statistics. Spatial statistics measured the TB incidence rate (TB patients per 100,000 residents) at the district level to determine its spatio-temporal distribution and to identify characteristics of change. Spatial autocorrelation analysis was used to detect global and local spatial autocorrelations across the study area. Purely spatial, purely temporal and space-time scan statistics were used to identify purely spatial, purely temporal and spatio-temporal clusters of TB at the district level. The other objective of this study was to compare the trajectory similarities between the incidence rates of TB and new smear-positive (NSP) TB patients in the resident population (NSPRP)/new smear-positive TB patients in the TB patient population (NSPTBP)/retreated smear-positive (RSP) TB patients in the resident population (RSPRP)/retreated smear-positive TB patients in the TB patient population (RSPTBP) to detect variations and loopholes in TB prevention and control among the districts in Beijing. The incidence rates in Beijing exhibited a gradual decrease from 2009 to 2014. Although global spatial autocorrelation was not detected overall across all of the districts of Beijing, individual districts did show evidence of local spatial autocorrelation: Chaoyang and Daxing were Low-Low districts over the six-year period. The purely spatial scan statistics analysis showed significant spatial clusters of high and low incidence rates; the purely temporal scan statistics showed the temporal cluster with a three-year period from 2009 to 2011 characterized by a high incidence rate; and the space-time scan statistics analysis showed significant spatio-temporal clusters. The distribution of the mean centres (MCs) showed that the general distributions of the NSPRP MCs and NSPTBP MCs were to the east of the incidence rate MCs. Conversely, the general distributions of the RSPRP MCs and the RSPTBP MCs were to the south of the incidence rate MCs. Based on the combined analysis of MC distribution characteristics and trajectory similarities, the NSP trajectory was most similar to the incidence rate trajectory. Thus, more attention should be focused on the discovery of NSP patients in the western part of Beijing, whereas the northern part of Beijing needs intensive treatment for RSP patients.

Evaluation of calling algorithms for array-CGH

Copy number variation (CNV) detection has become an integral part many of genetic studies and new technologies promise to revolutionize our ability to detect and link them to disease. However, recent studies highlight discrepancies in the genome wide CNV profile when measured by different technologies and even by the same technology. Furthermore, the change point algorithms used to call CNVs can have substantial disagreement on the same data set. We focus this article on comparative genomic hybridization (CGH) arrays because this platform lends itself well to accurate statistical modeling. We describe some newer methodological developments in local statistics that are well suited for CNV detection and calling on CGH arrays. Then we use both simulation studies and public data to compare these new local methods with the global methods that currently dominate literature. These results offer suggestions for choosing a particular method and provide insight to the lack of reproducibility that has been seen in the field so far.

Spatial Scan Statistics for Models with Excess Zeros and Overdispersion

Objective

To propose a more realistic model for disease cluster detection, through a modification of the spatial scan statistic to account simultaneously for inflated zeros and overdispersion.

Introduction

Spatial Scan Statistics [1] usually assume Poisson or Binomial distributed data, which is not adequate in many disease surveillance scenarios. For example, small areas distant from hospitals may exhibit a smaller number of cases than expected in those simple models. Also, underreporting may occur in underdeveloped regions, due to inefficient data collection or the difficulty to access remote sites. Those factors generate excess zero case counts or overdispersion, inducing a violation of the statistical model and also increasing the type I error (false alarms). Overdispersion occurs when data variance is greater than the predicted by the used model. To accommodate it, an extra parameter must be included; in the Poisson model, one makes the variance equal to the mean.

Methods

Tools like the Generalized Poisson (GP) and the Double Poisson [2] may be a better option for this kind of problem, modeling separately the mean and variance, which could be easily adjusted by covariates. When excess zeros occur, the Zero Inflated Poisson (ZIP) model is used, although ZIP’s estimated parameters may be severely biased if nonzero counts are too dispersed, compared to the Poisson distribution. In this case the Inflated Zero models for the Generalized Poisson (ZIGP), Double Poisson (ZIDP) and Negative Binomial (ZINB) could be good alternatives to the joint modeling of excess zeros and overdispersion. By one hand, Zero Inflated Poisson (ZIP) models were proposed using the spatial scan statistic to deal with the excess zeros [3]. By the other hand, another spatial scan statistic was based on a Poisson-Gamma mixture model for overdispersion [4]. In this work we present a model which includes inflated zeros and overdispersion simultaneously, based on the ZIDP model. Let the parameter p indicate the zero inflation. As the the remaining parameters of the observed cases map and the parameter p are not independent, the likelihood maximization process is not straightforward; it becomes even more complicated when we include covariates in the analysis. To solve this problem we introduce a vector of latent variables in order to factorize the likelihood, and obtain a facilitator for the maximization process using the E-M (Expectation-Maximization) algorithm. We derive the formulas to maximize iteratively the likelihood, and implement a computer program using the E-M algorithm to estimate the parameters under null and alternative hypothesis. The p-value is obtained via the Fast Double Bootstrap Test [5].

Results

Numerical simulations are conducted to assess the effectiveness of the method. We present results for Hanseniasis surveillance in the Brazilian Amazon in 2010 using this technique. We obtain the most likely spatial clusters for the Poisson, ZIP, Poisson-Gamma mixture and ZIDP models and compare the results.

Conclusions

The Zero Inflated Double Poisson Spatial Scan Statistic for disease cluster detection incorporates the flexibility of previous models, accounting for inflated zeros and overdispersion simultaneously.

The Hanseniasis study case map, due to excess of zero cases counts in many municipalities of the Brazilian Amazon and the presence of overdispersion, was a good benchmark to test the ZIDP model. The results obtained are easier to understand compared to each of the previous spatial scan statistic models, the Zero Inflated Poisson (ZIP) model and the Poisson-Gamma mixture model for overdispersion, taken separetely. The E-M algorithm and the Fast Double Bootstrap test are computationally efficient for this type of problem.

Fast Multidimensional Subset Scan for Outbreak Detection and Characterization

Objective

Introduction

Methods

Results

Conclusions

A scan statistic for identifying optimal risk windows in vaccine safety studies using self-controlled case series design

In the examination of the association between vaccines and rare adverse events after vaccination in postlicensure observational studies, it is challenging to define appropriate risk windows because prelicensure RCTs provide little insight on the timing of specific adverse events. Past vaccine safety studies have often used prespecified risk windows based on prior publications, biological understanding of the vaccine, and expert opinion. Recently, a data-driven approach was developed to identify appropriate risk windows for vaccine safety studies that use the self-controlled case series design. This approach employs both the maximum incidence rate ratio and the linear relation between the estimated incidence rate ratio and the inverse of average person time at risk, given a specified risk window. In this paper, we present a scan statistic that can identify appropriate risk windows in vaccine safety studies using the self-controlled case series design while taking into account the dependence of time intervals within an individual and while adjusting for time-varying covariates such as age and seasonality. This approach uses the maximum likelihood ratio test based on fixed-effects models, which has been used for analyzing data from self-controlled case series design in addition to conditional Poisson models.

Review of methods for space-time disease surveillance

A review of some methods for analysis of space-time disease surveillance data is presented. Increasingly, surveillance systems are capturing spatial and temporal data on disease and health outcomes in a variety of public health contexts. A vast and growing suite of methods exists for detection of outbreaks and trends in surveillance data and the selection of appropriate methods in a given surveillance context is not always clear. While most reviews of methods focus on algorithm performance, in practice, a variety of factors determine what methods are appropriate for surveillance. In this review, we focus on the role of contextual factors such as scale, scope, surveillance objective, disease characteristics, and technical issues in relation to commonly used approaches to surveillance. Methods are classified as testing-based or model-based approaches. Reviewing methods in the context of factors other than algorithm performance highlights important aspects of implementing and selecting appropriate disease surveillance methods.

Nonrandom clusters of palindromes in herpesvirus genomes

Palindromes are symmetrical words of DNA in the sense that they read exactly the same as their reverse complementary sequences. Representing the occurrences of palindromes in a DNA molecule as points on the unit interval, the scan statistics can be used to identify regions of unusually high concentration of palindromes. These regions have been associated with the replication origins on a few herpesviruses in previous studies. However, the use of scan statistics requires the assumption that the points representing the palindromes are independently and uniformly distributed on the unit interval. In this paper, we provide a mathematical basis for this assumption by showing that in randomly generated DNA sequences, the occurrences of palindromes can be approximated by a Poisson process. An easily computable upper bound on the Wasserstein distance between the palindrome process and the Poisson process is obtained. This bound is then used as a guide to choose an optimal palindrome length in the analysis of a collection of 16 herpesvirus genomes. Regions harboring significant palindrome clusters are identified and compared to known locations of replication origins. This analysis brings out a few interesting extensions of the scan statistics that can help formulate an algorithm for more accurate prediction of replication origins.

Biosurveillance applying scan statistics with multiple, disparate data sources

Researchers working on the Department of Defense Global Emerging Infections System (DoD-GEIS) pilot system, the Electronic Surveillance System for the Early Notification of Community-Based Epidemics (ESSENCE), have applied scan statistics for early outbreak detection using both traditional and nontraditional data sources. These sources include medical data indexed by International Classification of Disease, 9th Revision (ICD-9) diagnosis codes, as well as less-specific, but potentially timelier, indicators such as records of over-the-counter remedy sales and of school absenteeism. Early efforts employed the Kulldorff scan statistic as implemented in the SaTScan software of the National Cancer Institute. A key obstacle to this application is that the input data streams are typically based on time-varying factors, such as consumer behavior, rather than simply on the populations of the component subregions. We have used both modeling and recent historical data distributions to obtain background spatial distributions. Data analyses have provided guidance on how to condition and model input data to avoid excessive clustering. We have used this methodology in combining data sources for both retrospective studies of known outbreaks and surveillance of high-profile events of concern to local public health authorities. We have integrated the scan statistic capability into a Microsoft Access-based system in which we may include or exclude data sources, vary time windows separately for different data sources, censor data from subsets of individual providers or subregions, adjust the background computation method, and run retrospective or simulated studies.