Using laboratory-based surveillance data for prevention: an algorithm for detecting Salmonella outbreaks

Strengthening global health security through health early warning systems: A literature review and case study

Disease transmission is dependent on a variety of factors, including the characteristics of an event, such as crowding and shared accommodations, the potential of participants having prolonged exposure and close contact with infectious individuals, the type of activities, and the characteristics of the participants, such as their age and immunity to infectious agents [1-3]. Effective control of outbreaks of infectious diseases requires rapid diagnosis and intervention in high-risk settings. As a result, syndromic and event-based surveillance may be used to enhance the responsiveness of the surveillance system [1]. In public health, surveillance is collecting, analyzing, and interpreting data across time to inform decision-making and aid policy implementation [1]. In this review article we aimed to provide an overview of the principles, types, uses, advantages, and limitations of surveillance systems and to highlight the importance of early warning systems in response to the information received by disease surveillance. The study conducted a comprehensive literature search using several databases, selecting, and reviewing 78 articles that covered different types of surveillance systems, their applications, and their impact on controlling infectious diseases. The article also presents a case study from the Hajj gathering, which highlighted the development, evaluation, and impact of early warning systems on response to the information received by disease surveillance. The study concludes that ongoing disease surveillance should be accompanied by well-designed early warning and response systems, and continuous efforts should be invested in evaluating and validating these systems to minimize the risk of reporting delays and reducing the risk of outbreaks.

An Approach to Identifying Spatial Variability in Observed Infectious Disease Spread in a Prospective Time-Space Series with Applications to COVID-19 and Dengue Incidence

Most of the growing prospective analytic methods in space-time disease surveillance and intended functions of disease surveillance systems focus on earlier detection of disease outbreaks, disease clusters, or increased incidence. The spread of the virus such as SARS-CoV-2 has not been spatially and temporally uniform in an outbreak. With the identification of an infectious disease outbreak, recognizing and evaluating anomalies (excess and decline) of disease incidence spread at the time of occurrence during the course of an outbreak is a logical next step. We propose and formulate a hypergeometric probability model that investigates anomalies of infectious disease incidence spread at the time of occurrence in the timeline for many geographically described populations (e.g., hospitals, towns, counties) in an ongoing daily monitoring process. It is structured to determine whether the incidence grows or declines more rapidly in a region on the single current day or the most recent few days compared to the occurrence of the incidence during the previous few days relative to elsewhere in the surveillance period. The new method uses a time-varying baseline risk model, accounting for regularly (e.g., daily) updated information on disease incidence at the time of occurrence, and evaluates the probability of the deviation of particular frequencies to be attributed to sampling fluctuations, accounting for the unequal variances of the rates due to different population bases in geographical units. We attempt to present and illustrate a new model to advance the investigation of anomalies of infectious disease incidence spread by analyzing subsamples of spatiotemporal disease surveillance data from Taiwan on dengue and COVID-19 incidence which are mosquito-borne and contagious infectious diseases, respectively. Efficient R programs for computation are available to implement the two approximate formulae of the hypergeometric probability model for large numbers of events.

Evaluation of the application of sequence data to the identification of outbreaks of disease using anomaly detection methods

Anomaly detection methods have a great potential to assist the detection of diseases in animal production systems. We used sequence data of Porcine Reproductive and Respiratory Syndrome (PRRS) to define the emergence of new strains at the farm level. We evaluated the performance of 24 anomaly detection methods based on machine learning, regression, time series techniques and control charts to identify outbreaks in time series of new strains and compared the best methods using different time series: PCR positives, PCR requests and laboratory requests. We introduced synthetic outbreaks of different size and calculated the probability of detection of outbreaks (POD), sensitivity (Se), probability of detection of outbreaks in the first week of appearance (POD1w) and background alarm rate (BAR). The use of time series of new strains from sequence data outperformed the other types of data but POD, Se, POD1w were only high when outbreaks were large. The methods based on Long Short-Term Memory (LSTM) and Bayesian approaches presented the best performance. Using anomaly detection methods with sequence data may help to identify the emergency of cases in multiple farms, but more work is required to improve the detection with time series of high variability. Our results suggest a promising application of sequence data for early detection of diseases at a production system level. This may provide a simple way to extract additional value from routine laboratory analysis. Next steps should include validation of this approach in different settings and with different diseases.

Real-time risk ranking of emerging epidemics based on optimized moving average prediction limit-taking the COVID-19 pandemic as an example

BACKGROUND

Mathematical models to forecast the risk trend of the COVID-19 pandemic timely are of great significance to control the pandemic, but the requirement of manual operation and many parameters hinders their efficiency and value for application. This study aimed to establish a convenient and prompt one for monitoring emerging infectious diseases online and achieving risk assessment in real time.

METHODS

The Optimized Moving Average Prediction Limit (Op-MAPL) algorithm model analysed real-time COVID-19 data online and was validated using the data of the Delta variant in India and the Omicron in the United States. Then, the model was utilized to determine the infection risk level of the Omicron in Shanghai and Beijing.

RESULTS

The Op-MAPL model can predict the epidemic peak accurately. The daily risk ranking was stable and predictive, with an average accuracy of 87.85% within next 7 days. Early warning signals were issued for Shanghai and Beijing on February 28 and April 23, 2022, respectively. The two cities were rated as medium-high risk or above from March 27 to April 20 and from April 24 to May 5, indicating that the pandemic had entered a period of rapid increase. After April 21 and May 26, the risk level was downgraded to medium and became stable by the algorithm, indicating that the pandemic had been controlled well and mitigated gradually.

CONCLUSIONS

The Op-MAPL relies on nothing but an indicator to assess the risk level of the COVID-19 pandemic with different data sources and granularities. This forward-looking method realizes real-time monitoring and early warning effectively to provide a valuable reference to prevent and control infectious diseases.

Epidemiological cluster identification using multiple data sources: an approach using logistic regression

In the management of infectious disease outbreaks, grouping cases into clusters and understanding their underlying epidemiology are fundamental tasks. In genomic epidemiology, clusters are typically identified either using pathogen sequences alone or with sequences in combination with epidemiological data such as location and time of collection. However, it may not be feasible to culture and sequence all pathogen isolates, so sequence data may not be available for all cases. This presents challenges for identifying clusters and understanding epidemiology, because these cases may be important for transmission. Demographic, clinical and location data are likely to be available for unsequenced cases, and comprise partial information about their clustering. Here, we use statistical modelling to assign unsequenced cases to clusters already identified by genomic methods, assuming that a more direct method of linking individuals, such as contact tracing, is not available. We build our model on pairwise similarity between cases to predict whether cases cluster together, in contrast to using individual case data to predict the cases' clusters. We then develop methods that allow us to determine whether a pair of unsequenced cases are likely to cluster together, to group them into their most probable clusters, to identify which are most likely to be members of a specific (known) cluster, and to estimate the true size of a known cluster given a set of unsequenced cases. We apply our method to tuberculosis data from Valencia, Spain. Among other applications, we find that clustering can be predicted successfully using spatial distance between cases and whether nationality is the same. We can identify the correct cluster for an unsequenced case, among 38 possible clusters, with an accuracy of approximately 35 %, higher than both direct multinomial regression (17 %) and random selection (< 5 %).

The detection of the epidemic phase of COVID-19 and the timing of social distancing policies in Korea

Objectives

Observing cumulative and new daily confirmed cases of COVID-19, disease control authorities respond to a surge in cases with social distancing measures or economic lockdown. The question in this article is whether we can gather more useful information from a readily available time series data set of day-to-day changes in confirmed cases of COVID-19.

Study design

Time-series data analysis was done using a hidden Markov model.

Methods

Day-to-day differences in confirmed cases of COVID-19 in Korea from February 19, 2020, to July 13, 2021, were modeled via a hidden Markov model. The results from the model were compared with the effective reproduction number and the Korean government's response.

Results

The model reports that Korea was in an epidemic phase from August 2020 and from mid-November 2020, the second and third epidemic waves. The government's response, represented by the Government Response Stringency Index, was not timely during the epidemic phases. The results from the model may also be more helpful to detect the onset of the epidemic phase of an infectious disease than the effective reproduction number.

Conclusions

The model can reveal a hidden epidemic phase and help disease control authorities to respond more promptly and effectively.

An aberration detection-based approach for sentinel syndromic surveillance of COVID-19 and other novel influenza-like illnesses

Coronavirus Disease 2019 has emerged as a significant global concern, triggering harsh public health restrictions in a successful bid to curb its exponential growth. As discussion shifts towards relaxation of these restrictions, there is significant concern of second-wave resurgence. The key to managing these outbreaks is early detection and intervention, and yet there is a significant lag time associated with usage of laboratory confirmed cases for surveillance purposes. To address this, syndromic surveillance can be considered to provide a timelier alternative for first-line screening. Existing syndromic surveillance solutions are however typically focused around a known disease and have limited capability to distinguish between outbreaks of individual diseases sharing similar syndromes. This poses a challenge for surveillance of COVID-19 as its active periods tend to overlap temporally with other influenza-like illnesses. In this study we explore performing sentinel syndromic surveillance for COVID-19 and other influenza-like illnesses using a deep learning-based approach. Our methods are based on aberration detection utilizing autoencoders that leverages symptom prevalence distributions to distinguish outbreaks of two ongoing diseases that share similar syndromes, even if they occur concurrently. We first demonstrate that this approach works for detection of outbreaks of influenza, which has known temporal boundaries. We then demonstrate that the autoencoder can be trained to not alert on known and well-managed influenza-like illnesses such as the common cold and influenza. Finally, we applied our approach to 2019-2020 data in the context of a COVID-19 syndromic surveillance task to demonstrate how implementation of such a system could have provided early warning of an outbreak of a novel influenza-like illness that did not match the symptom prevalence profile of influenza and other known influenza-like illnesses.

Investigations of Possible Multistate Outbreaks of Salmonella, Shiga Toxin-Producing Escherichia coli, and Listeria monocytogenes Infections - United States, 2016

PROBLEM/CONDITION

Salmonella, Shiga toxin-producing Escherichia coli (STEC), and Listeria monocytogenes are the leading causes of multistate foodborne disease outbreaks in the United States. Responding to multistate outbreaks quickly and effectively and applying lessons learned about outbreak sources, modes of transmission, and risk factors for infection can prevent additional outbreak-associated illnesses and save lives. This report summarizes the investigations of multistate outbreaks and possible outbreaks of Salmonella, STEC, and L. monocytogenes infections coordinated by CDC during the 2016 reporting period.

PERIOD COVERED

2016. An investigation was considered to have occurred in 2016 if it began during 2016 and ended on or before March 31, 2017, or if it began before January 1, 2016, and ended during March 31, 2016-March 31, 2017.

DESCRIPTION OF SYSTEM

CDC maintains a database of investigations of possible multistate foodborne and animal-contact outbreaks caused by Salmonella, STEC, and L. monocytogenes. Data were collected by local, state, and federal investigators during the detection, investigation and response, and control phases of the outbreak investigations. Additional data sources used for this report included PulseNet, the national molecular subtyping network based on isolates uploaded by local, state, and federal laboratories, and the Foodborne Disease Outbreak Surveillance System (FDOSS), which collects information from state, local, and territorial health departments and federal agencies about single-state and multistate foodborne disease outbreaks in the United States. Multistate outbreaks reported to FDOSS were linked using a unique outbreak identifier to obtain food category information when a confirmed or suspected food source was identified. Food categories were determined and assigned in FDOSS according to a classification scheme developed by CDC, the Food and Drug Administration (FDA), and the U.S. Department of Agriculture Food Safety and Inspection Service (FSIS) in the Interagency Food Safety Analytics Collaboration. A possible multistate outbreak was determined by expert judgment to be an outbreak if supporting data (e.g., temporal, geographic, demographic, dietary, travel, or food history) suggested a common source. A solved outbreak was an outbreak for which a specific kind of food or animal was implicated (i.e., confirmed or suspected) as the source. Outbreak-level variables included number of illnesses, hospitalizations, cases of hemolytic uremic syndrome (HUS), and deaths; the number of states with illnesses; date of isolation for the earliest and last cases; demographic data describing patients associated with a possible outbreak (e.g., age, sex, and state of residence); the types of data collected (i.e., epidemiologic, traceback, or laboratory); the outbreak source, mode of transmission, and exposure location; the name or brand of the source; whether the source was suspected or confirmed; whether a food was imported into the United States; the types of regulatory agencies involved; whether regulatory action was taken (and what type of action); whether an outbreak was publicly announced by CDC via website posting; beginning and end date of the investigation; and general comments about the investigation. The number of illnesses, hospitalizations, cases of HUS, and deaths were characterized by transmission mode, pathogen, outcome (i.e., unsolved, solved with suspected source, or solved with confirmed source), source, and food or animal category.

RESULTS

During the 2016 reporting period, 230 possible multistate outbreaks were detected and 174 were investigated. A median of 24 possible outbreaks was under investigation per week, and investigations were open for a median of 37 days. Of these 174 possible outbreaks investigated, 56 were excluded from this analysis because they occurred in a single state, were linked to international travel, or were pseudo-outbreaks (e.g., a group of similar isolates resulting from laboratory media contamination rather than infection in patients). Of the remaining 118 possible multistate outbreaks, 50 were determined to be outbreaks and 39 were solved (18 with a confirmed food source, 10 with a suspected food source, 10 with a confirmed animal source, and one with a suspected animal source). Sprouts were the most commonly implicated food category in solved multistate foodborne outbreaks (five). Chicken was the source of the most foodborne outbreak-related illnesses (134). Three outbreaks involved novel food-pathogen pairs: flour and STEC, frozen vegetables and L. monocytogenes, and bagged salad and L. monocytogenes. Eleven outbreaks were attributed to contact with animals (10 attributed to contact with backyard poultry and one to small turtles). Thirteen of 18 multistate foodborne disease outbreaks with confirmed sources resulted in product action, including 10 outbreaks with recalls, two with market withdrawals, and one with an FSIS public health alert. Twenty outbreaks, including 11 foodborne and nine animal-contact outbreaks, were announced to the public by CDC via its website, Facebook, and Twitter. These announcements resulted in approximately 910,000 webpage views, 55,000 likes, 66,000 shares, and 5,800 retweets.

INTERPRETATION

During the 2016 reporting period, investigations of possible multistate outbreaks occurred frequently, were resource intensive, and required a median of 37 days of investigation. Fewer than half (42%) of the 118 possible outbreaks investigated were determined to have sufficient data to meet the definition of a multistate outbreak. Moreover, of the 50 outbreaks with sufficient data, approximately three fourths were solved.

PUBLIC HEALTH ACTION

Close collaboration among CDC, FDA, FSIS and state and local health and agriculture partners is central to successful outbreak investigations. Identification of novel outbreak sources and trends in sources provides insights into gaps in food safety and safe handling of animals, which helps focus prevention strategies. Summarizing investigations of possible multistate outbreaks can provide insights into the investigative process, improve future investigations, and help prevent illnesses. Although identifying and investigating possible multistate outbreaks require substantial resources and investment in public health infrastructure, they are important in determining outbreak sources and implementing prevention and control measures.

Visual tools to assess the plausibility of algorithm-identified infectious disease clusters: an application to mumps data from the Netherlands dating from January 2009 to June 2016

Introduction

With growing amounts of data available, identification of clusters of persons linked to each other by transmission of an infectious disease increasingly relies on automated algorithms. We propose cluster finding to be a two-step process: first, possible transmission clusters are identified using a cluster algorithm, second, the plausibility that the identified clusters represent genuine transmission clusters is evaluated.

Aim

To introduce visual tools to assess automatically identified clusters.

Methods

We developed tools to visualise: (i) clusters found in dimensions of time, geographical location and genetic data; (ii) nested sub-clusters within identified clusters; (iii) intra-cluster pairwise dissimilarities per dimension; (iv) intra-cluster correlation between dimensions. We applied our tools to notified mumps cases in the Netherlands with available disease onset date (January 2009 – June 2016), geographical information (location of residence), and pathogen sequence data (n = 112). We compared identified clusters to clusters reported by the Netherlands Early Warning Committee (NEWC).

Results

We identified five mumps clusters. Three clusters were considered plausible. One was questionable because, in phylogenetic analysis, genetic sequences related to it segregated in two groups. One was implausible with no smaller nested clusters, high intra-cluster dissimilarities on all dimensions, and low intra-cluster correlation between dimensions. The NEWC reports concurred with our findings: the plausible/questionable clusters corresponded to reported outbreaks; the implausible cluster did not.

Conclusion

Our tools for assessing automatically identified clusters allow outbreak investigators to rapidly spot plausible transmission clusters for mumps and other human-to-human transmissible diseases. This fast information processing potentially reduces workload.

Use of the CUSUM and Shewhart control chart methods to identify changes of public health significance using childhood blood lead surveillance data

Supplemental Digital Content is available in the text.

Local, state, and national childhood blood lead surveillance is based on healthcare providers and clinical laboratories reporting test results to public health departments. Increased interest in detecting blood lead level (BLL) patterns and changes of potential public health significance in a timely manner has highlighted the need for surveillance systems to rapidly detect and investigate these events.

A real-time spatio-temporal syndromic surveillance system with application to small companion animals

Lack of disease surveillance in small companion animals worldwide has contributed to a deficit in our ability to detect and respond to outbreaks. In this paper we describe the first real-time syndromic surveillance system that conducts integrated spatio-temporal analysis of data from a national network of veterinary premises for the early detection of disease outbreaks in small animals. We illustrate the system’s performance using data relating to gastrointestinal disease in dogs and cats. The data consist of approximately one million electronic health records for dogs and cats, collected from 458 UK veterinary premises between March 2014 and 2016. For this illustration, the system predicts the relative reporting rate of gastrointestinal disease amongst all presentations, and updates its predictions as new data accrue. The system was able to detect simulated outbreaks of varying spatial geometry, extent and severity. The system is flexible: it generates outcomes that are easily interpretable; the user can set their own outbreak detection thresholds. The system provides the foundation for prompt detection and control of health threats in companion animals.

Statistical outbreak detection by joining medical records and pathogen similarity

We present a statistical inference model for the detection and characterization of outbreaks of hospital associated infection. The approach combines patient exposures, determined from electronic medical records, and pathogen similarity, determined by whole-genome sequencing, to simultaneously identify probable outbreaks and their root-causes. We show how our model can be used to target isolates for whole-genome sequencing, improving outbreak detection and characterization even without comprehensive sequencing. Additionally, we demonstrate how to learn model parameters from reference data of known outbreaks. We demonstrate model performance using semi-synthetic experiments.

A graph-based evidence synthesis approach to detecting outbreak clusters: An application to dog rabies

Early assessment of infectious disease outbreaks is key to implementing timely and effective control measures. In particular, rapidly recognising whether infected individuals stem from a single outbreak sustained by local transmission, or from repeated introductions, is crucial to adopt effective interventions. In this study, we introduce a new framework for combining several data streams, e.g. temporal, spatial and genetic data, to identify clusters of related cases of an infectious disease. Our method explicitly accounts for underreporting, and allows incorporating preexisting information about the disease, such as its serial interval, spatial kernel, and mutation rate. We define, for each data stream, a graph connecting all cases, with edges weighted by the corresponding pairwise distance between cases. Each graph is then pruned by removing distances greater than a given cutoff, defined based on preexisting information on the disease and assumptions on the reporting rate. The pruned graphs corresponding to different data streams are then merged by intersection to combine all data types; connected components define clusters of cases related for all types of data. Estimates of the reproduction number (the average number of secondary cases infected by an infectious individual in a large population), and the rate of importation of the disease into the population, are also derived. We test our approach on simulated data and illustrate it using data on dog rabies in Central African Republic. We show that the outbreak clusters identified using our method are consistent with structures previously identified by more complex, computationally intensive approaches.

Early assessment of infectious disease outbreaks is key to implementing timely and effective control measures. In particular, rapidly recognising whether infected individuals stem from a single outbreak sustained by local transmission, or from repeated introductions, is crucial to adopt effective interventions. In this study, we introduce a new approach which combines different types of data to identify clusters of related cases of an infectious disease. This approach relies on representing each type of data (e.g. temporal, spatial, or genetic) as a graph where nodes are cases, and two nodes are connected if the corresponding cases are closely related for this data. Our method then identifies clusters of cases which likely stem from the same introduction. Furthermore, we can use the size of these clusters to infer transmissibility of the disease and the number of importations of the pathogen into the population. We apply this approach to analyse dog rabies epidemics in Central African Republic. We show that outbreak clusters identified using our method are consistent with structures previously identified by more complex and computationally intensive approaches. Using simulated rabies epidemics, we show that our method has excellent potential for optimally detecting outbreak clusters. We also identify promising areas of research for transforming our method into a routine analysis tool for processing disease surveillance data.

Sounding the alarm: Defining thresholds to trigger a public health response to monkeypox

Endemic to the Democratic Republic of the Congo (DRC), monkeypox is a zoonotic disease that causes smallpox-like illness in humans. Observed fluctuations in reported cases over time raises questions about when it is appropriate to mount a public health response, and what specific actions should be taken. We evaluated three different thresholds to differentiate between baseline and heightened disease incidence, and propose a novel, tiered algorithm for public health action. Monkeypox surveillance data from Tshuapa Province, 2011-2013, were used to calculate three different statistical thresholds: Cullen, c-sum, and a World Health Organization (WHO) method based on monthly incidence. When the observed cases exceeded the threshold for a given month, that month was considered to be 'aberrant'. For each approach, the number of aberrant months detected was summed by year-each method produced vastly different results. The Cullen approach generated a number of aberrant signals over the period of consideration (9/36 months). The c-sum method was the most sensitive (30/36 months), followed by the WHO method (12/24 months). We conclude that triggering public health action based on signals detected by a single method may be inefficient and overly simplistic for monkeypox. We propose instead a response algorithm that integrates an objective threshold (WHO method) with contextual information about epidemiological and spatiotemporal links between suspected cases to determine whether a response should be operating under i) routine surveillance ii) alert status, or iii) outbreak status. This framework could be modified and adopted by national and zone level health workers in monkeypox-endemic countries. Lastly, we discuss considerations for selecting thresholds for monkeypox outbreaks across gradients of endemicity and public health resources.

Global Research on Syndromic Surveillance from 1993 to 2017: Bibliometric Analysis and Visualization

Syndromic Surveillance aims at analyzing medical data to detect clusters of illness or forecast disease outbreaks. Although the research in this field is flourishing in terms of publications, an insight of the global research output has been overlooked. This paper aims at analyzing the global scientific output of the research from 1993 to 2017. To this end, the paper uses bibliometric analysis and visualization to achieve its goal. Particularly, a data processing framework was proposed based on citation datasets collected from Scopus and Clarivate Analytics’ Web of Science Core Collection (WoSCC). The bibliometric method and Citespace were used to analyze the institutions, countries, and research areas as well as the current hotspots and trends. The preprocessed dataset includes 14,680 citation records. The analysis uncovered USA, England, Canada, France and Australia as the top five most productive countries publishing about Syndromic Surveillance. On the other hand, at the Pinnacle of academic institutions are the US Centers for Disease Control and Prevention (CDC). The reference co-citation analysis uncovered the common research venues and further analysis of the keyword cooccurrence revealed the most trending topics. The findings of this research will help in enriching the field with a comprehensive view of the status and future trends of the research on Syndromic Surveillance.

Choosing the best algorithm for event detection based on the intended application: A conceptual framework for syndromic surveillance

There is an extensive list of methods available for the early detection of an epidemic signal in syndromic surveillance data. However, there is no commonly accepted classification system for the statistical methods used for event detection in syndromic surveillance. Comparing and choosing appropriate event detection algorithms is an increasingly challenging task. Although lists of selection criteria, and statistical methods used for signal detection have been reported, selection criteria are rarely linked to a specific set of appropriate statistical methods. The paper presents a practical approach for guiding surveillance practitioners to make an informed choice from among the most popular event detection algorithms based on the intended application of the algorithm. We developed selection criteria by mapping the assumptions and performance characteristics of event detection algorithms directly to important characteristics of the time series used in syndromic surveillance. We also considered types of epidemics that may be expected and other characteristics of the surveillance system. These guidelines will provide decisions makers, data analysts, public health practitioners, and researchers with a comprehensive but practical overview of the domain, which may reduce the technical barriers to the development and implementation of syndromic surveillance systems in animal and human health. The classification scheme was restricted to univariate and temporal methods because they are the most commonly used algorithms in syndromic surveillance.

Assessing current temporal and space-time anomalies of disease incidence

Approaches used to early and accurately characterize epidemiologic patterns of disease incidence in a temporal and spatial series are becoming increasingly important. Cluster tests are generally designed for retrospective detection of epidemiologic anomalies in a temporal or space-time series. Timely identification of anomalies of disease or poisoning incidence during ongoing surveillance or an outbreak requires the use of sensitive statistical methods that recognize an incidence pattern at the time of occurrence. This report describes 2 novel analytical methods that focus on detecting anomalies of incidence at the time of occurrence in a temporal and space-time series. The first method describes the paucity of incidence at the time of occurrence in an ongoing surveillance and is designed to evaluate whether a decline in incidence occurs on the single current day or during the most recent few days. The second method provides an overall assessment of current clustering or paucity of incidence in a space-time series, allowing for several space regions. We illustrate the application of these methods using a subsample of a temporal series of data on the largest dengue outbreak in Taiwan in 2015 since World War II and demonstrate that they are useful to efficiently monitor incoming data for current clustering and paucity of incidence in a temporal and space-time series. In light of the recent global emergence and resurgence of Zika, dengue, and chikungunya infection, these approaching for detecting current anomalies of incidence in the ongoing surveillance of disease are particularly desired and needed.

The Validation and Implications of Using Whole Genome Sequencing as a Replacement for Traditional Serotyping for a National Salmonella Reference Laboratory

Salmonella serotyping remains the gold-standard tool for the classification of Salmonella isolates and forms the basis of Canada’s national surveillance program for this priority foodborne pathogen. Public health officials have been increasingly looking toward whole genome sequencing (WGS) to provide a large set of data from which all the relevant information about an isolate can be mined. However, rigorous validation and careful consideration of potential implications in the replacement of traditional surveillance methodologies with WGS data analysis tools is needed. Two in silico tools for Salmonella serotyping have been developed, the Salmonella in silico Typing Resource (SISTR) and SeqSero, while seven gene MLST for serovar prediction can be adapted for in silico analysis. All three analysis methods were assessed and compared to traditional serotyping techniques using a set of 813 verified clinical and laboratory isolates, including 492 Canadian clinical isolates and 321 isolates of human and non-human sources. Successful results were obtained for 94.8, 88.2, and 88.3% of the isolates tested using SISTR, SeqSero, and MLST, respectively, indicating all would be suitable for maintaining historical records, surveillance systems, and communication structures currently in place and the choice of the platform used will ultimately depend on the users need. Results also pointed to the need to reframe serotyping in the genomic era as a test to understand the genes that are carried by an isolate, one which is not necessarily congruent with what is antigenically expressed. The adoption of WGS for serotyping will provide the simultaneous collection of information that can be used by multiple programs within the current surveillance paradigm; however, this does not negate the importance of the various programs or the role of serotyping going forward.

Monitoring antimicrobial resistance (AMR) using CUSUM control charts

We evaluated the use of the Cumulative Summation (CUSUM) control chart methodology for detection of an excessive increase in antimicrobial-resistant (AMR) bacteria acquisition. We used administrative, clinical and bacteriological data from all 157,570 patients hospitalized for at least 48 h from January 1, 2010 to December 31, 2015 in a 654-bed university teaching hospital in Paris, France. Monthly computed CUSUM were evaluated for the detection of out-of-control situations, defined as incidence rates of acquired AMR bacterial colonization exceeding acceptable thresholds at the hospital and ward levels (based on six selected wards) for AMR bacteria overall and Extended-spectrum beta-lactamases Enterobacteriaceae (ESBL-E) and Methicillin-resistant Staphylococcus aureus (MRSA), specifically. During the study period, 1,403 samples of acquired AMR bacteria were identified including 1,129 ESBL-E and 151 MRSA. The incidence rate of acquired AMR bacteria was stable at the hospital and the wards level. When based on AMR bacteria overall, CUSUM alarms were triggered at the hospital level and at the ward level in four units. For ESBL-E, CUSUM tests generated alarms at the hospital level and for the same four wards, and for MRSA, CUSUM tests detected out-of-control situations in all the wards. The CUSUM approach appears complementary with hospital infection control strategies currently in practice and appears of interest in common practice as a simple tool for AMR surveillance.

National Outbreak of Multidrug Resistant Salmonella Heidelberg Infections Linked to a Single Poultry Company

IMPORTANCE

This large outbreak of foodborne salmonellosis demonstrated the complexity of investigating outbreaks linked to poultry products. The outbreak also highlighted the importance of efforts to strengthen food safety policies related to Salmonella in chicken parts and has implications for future changes within the poultry industry.

OBJECTIVE

To investigate a large multistate outbreak of multidrug resistant Salmonella Heidelberg infections.

DESIGN

Epidemiologic and laboratory investigations of patients infected with the outbreak strains of Salmonella Heidelberg and traceback of possible food exposures.

SETTING

United States. Outbreak period was March 1, 2013 through July 11, 2014.

PATIENTS

A case was defined as illness in a person infected with a laboratory-confirmed Salmonella Heidelberg with 1 of 7 outbreak pulsed-field gel electrophoresis (PFGE) XbaI patterns with illness onset from March 1, 2013 through July 11, 2014. A total of 634 case-patients were identified through passive surveillance; 200/528 (38%) were hospitalized, none died.

RESULTS

Interviews were conducted with 435 case-patients: 371 (85%) reported eating any chicken in the 7 days before becoming ill. Of 273 case-patients interviewed with a focused questionnaire, 201 (74%) reported eating chicken prepared at home. Among case-patients with available brand information, 152 (87%) of 175 patients reported consuming Company A brand chicken. Antimicrobial susceptibility testing was completed on 69 clinical isolates collected from case-patients; 67% were drug resistant, including 24 isolates (35%) that were multidrug resistant. The source of Company A brand chicken consumed by case-patients was traced back to 3 California production establishments from which 6 of 7 outbreak strains were isolated.

CONCLUSIONS

Epidemiologic, laboratory, traceback, and environmental investigations conducted by local, state, and federal public health and regulatory officials indicated that consumption of Company A chicken was the cause of this outbreak. The outbreak involved multiple PFGE patterns, a variety of chicken products, and 3 production establishments, suggesting a reservoir for contamination upstream from the production establishments. Sources of bacteria and genes responsible for resistance, such as farms providing birds for slaughter or environmental reservoir on farms that raise chickens, might explain how multiple PFGE patterns were linked to chicken from 3 separate production establishments and many different poultry products.

Refining historical limits method to improve disease cluster detection, New York City, New York, USA

Our refinements corrected for major biases, preserved simplicity, and improved validity.

Since the early 2000s, the Bureau of Communicable Disease of the New York City Department of Health and Mental Hygiene has analyzed reportable infectious disease data weekly by using the historical limits method to detect unusual clusters that could represent outbreaks. This method typically produced too many signals for each to be investigated with available resources while possibly failing to signal during true disease outbreaks. We made method refinements that improved the consistency of case inclusion criteria and accounted for data lags and trends and aberrations in historical data. During a 12-week period in 2013, we prospectively assessed these refinements using actual surveillance data. The refined method yielded 74 signals, a 45% decrease from what the original method would have produced. Fewer and less biased signals included a true citywide increase in legionellosis and a localized campylobacteriosis cluster subsequently linked to live-poultry markets. Future evaluations using simulated data could complement this descriptive assessment.

Evaluation of U.S. Poison Center Data for Surveillance of Foodborne Disease

INTRODUCTION

Enhancing foodborne disease (FBD) surveillance and improving the timeliness of outbreak detection have been identified as public health priorities. Consumer complaint data have become increasingly useful for FBD surveillance and the detection of outbreaks. Calls to poison centers are a potential source of consumer complaint data. A retrospective analysis of data from the National Poison Data System (NPDS) (2000-2011) was undertaken to evaluate the value of data collected through the United States poison centers for detection of large national outbreaks and recalls.

METHODS

Demographic and clinical data were summarized. Prevalences of FBD calls were calculated and analyzed for time trends. Significant increases in daily call prevalences were identified, and dates of the increases were compared to the announcement of 18 national outbreaks/recalls.

RESULTS

Over the 12-year period, there were 433,788 unique calls self-reporting a suspected FBD exposure in humans. Overall, daily call prevalences decreased over time. Only about half of callers reported common gastrointestinal clinical effects. Of the 42 identified significant increases in call prevalences, none occurred within 14 days before an outbreak announcement; 7 occurred within 14 days after an outbreak announcement.

CONCLUSIONS

Based on this analysis, there are significant limitations to using self-reported FBD exposures to NPDS as a source of information for FBD surveillance of large national outbreaks and recalls; however, a syndromic approach may yield different results and should be explored. Improved data collection and coordination with public health agencies may improve the ability to use NPDS data to monitor FBD in near real-time, identify potential outbreaks, and improve situational awareness.

Using an adjusted Serfling regression model to improve the early warning at the arrival of peak timing of influenza in Beijing

Serfling-type periodic regression models have been widely used to identify and analyse epidemic of influenza. In these approaches, the baseline is traditionally determined using cleaned historical non-epidemic data. However, we found that the previous exclusion of epidemic seasons was empirical, since year-year variations in the seasonal pattern of activity had been ignored. Therefore, excluding fixed ‘epidemic’ months did not seem reasonable. We made some adjustments in the rule of epidemic-period removal to avoid potentially subjective definition of the start and end of epidemic periods. We fitted the baseline iteratively. Firstly, we established a Serfling regression model based on the actual observations without any removals. After that, instead of manually excluding a predefined ‘epidemic’ period (the traditional method), we excluded observations which exceeded a calculated boundary. We then established Serfling regression once more using the cleaned data and excluded observations which exceeded a calculated boundary. We repeated this process until the R² value stopped to increase. In addition, the definitions of the onset of influenza epidemic were heterogeneous, which might make it impossible to accurately evaluate the performance of alternative approaches. We then used this modified model to detect the peak timing of influenza instead of the onset of epidemic and compared this model with traditional Serfling models using observed weekly case counts of influenza-like illness (ILIs), in terms of sensitivity, specificity and lead time. A better performance was observed. In summary, we provide an adjusted Serfling model which may have improved performance over traditional models in early warning at arrival of peak timing of influenza.

Forecasting influenza epidemics from multi-stream surveillance data in a subtropical city of China

Background

Influenza has been associated with heavy burden of mortality and morbidity in subtropical regions. However, timely forecast of influenza epidemic in these regions has been hindered by unclear seasonality of influenza viruses. In this study, we developed a forecasting model by integrating multiple sentinel surveillance data to predict influenza epidemics in a subtropical city Shenzhen, China.

Methods

Dynamic linear models with the predictors of single or multiple surveillance data for influenza-like illness (ILI) were adopted to forecast influenza epidemics from 2006 to 2012 in Shenzhen. Temporal coherence of these surveillance data with laboratory-confirmed influenza cases was evaluated by wavelet analysis and only the coherent data streams were entered into the model. Timeliness, sensitivity and specificity of these models were also evaluated to compare their performance.

Results

Both influenza virology data and ILI consultation rates in Shenzhen demonstrated a significant annual seasonal cycle (p<0.05) during the entire study period, with occasional deviations observed in some data streams. The forecasting models that combined multi-stream ILI surveillance data generally outperformed the models with single-stream ILI data, by providing more timely, sensitive and specific alerts.

Conclusions

Forecasting models that combine multiple sentinel surveillance data can be considered to generate timely alerts for influenza epidemics in subtropical regions like Shenzhen.

Determination of French influenza outbreaks periods between 1985 and 2011 through a web-based Delphi method

BACKGROUND

Assessing the accuracy of influenza epidemic periods determined by statistical models is important to improve the performance of algorithms used in real-time syndromic surveillance systems. This is a difficult problem to address in the absence of a reliable gold standard. The objective of this study is to establish an expert-based determination of the start and the end of influenza epidemics in France.

METHODS

A three-round international web-based Delphi survey was proposed to 288 eligible influenza experts. Fifty-seven (20%) experts completed the three-rounds of the study. The experts were invited to indicate the starting and the ending week of influenza epidemics, on 32 time-series graphs of influenza seasons drawn using data from the French Sentinelles Network (Influenza-like illness incidence rates) and virological data from the WHO-FluNet. Twenty-six of 32 time-series graphs proposed corresponded to each of the French influenza seasons observed between 1985 and 2011. Six influenza seasons were proposed twice at each round to measure variation among expert responses.

RESULTS

We obtained consensual results for 88% (23/26) of the epidemic periods. In two or three rounds (depending on the season) answers gathered around modes, and the internal control demonstrated a good reproducibility of the answers. Virological data did not appear to have a significant impact on the answers or the level of consensus, except for a season with a major mismatch between virological and incidence data timings.

CONCLUSIONS

Thanks to this international web-based Delphi survey, we obtained reproducible, stable and consensual results for the majority of the French influenza epidemic curves analysed. The detailed curves together with the estimates from the Delphi study could be a helpful tool for assessing the performance of statistical outbreak detection methods, in order to optimize them.

Laboratory-based prospective surveillance for community outbreaks of Shigella spp. in Argentina

BACKGROUND

To implement effective control measures, timely outbreak detection is essential. Shigella is the most common cause of bacterial diarrhea in Argentina. Highly resistant clones of Shigella have emerged, and outbreaks have been recognized in closed settings and in whole communities. We hereby report our experience with an evolving, integrated, laboratory-based, near real-time surveillance system operating in six contiguous provinces of Argentina during April 2009 to March 2012.

METHODOLOGY

To detect localized shigellosis outbreaks timely, we used the prospective space-time permutation scan statistic algorithm of SaTScan, embedded in WHONET software. Twenty three laboratories sent updated Shigella data on a weekly basis to the National Reference Laboratory. Cluster detection analysis was performed at several taxonomic levels: for all Shigella spp., for serotypes within species and for antimicrobial resistance phenotypes within species. Shigella isolates associated with statistically significant signals (clusters in time/space with recurrence interval ≥365 days) were subtyped by pulsed field gel electrophoresis (PFGE) using PulseNet protocols.

PRINCIPAL FINDINGS

In three years of active surveillance, our system detected 32 statistically significant events, 26 of them identified before hospital staff was aware of any unexpected increase in the number of Shigella isolates. Twenty-six signals were investigated by PFGE, which confirmed a close relationship among the isolates for 22 events (84.6%). Seven events were investigated epidemiologically, which revealed links among the patients. Seventeen events were found at the resistance profile level. The system detected events of public health importance: infrequent resistance profiles, long-lasting and/or re-emergent clusters and events important for their duration or size, which were reported to local public health authorities.

CONCLUSIONS/SIGNIFICANCE

The WHONET-SaTScan system may serve as a model for surveillance and can be applied to other pathogens, implemented by other networks, and scaled up to national and international levels for early detection and control of outbreaks.

Validity and timeliness of syndromic influenza surveillance during the autumn/winter wave of A (H1N1) influenza 2009: results of emergency medical dispatch, ambulance and emergency department data from three European regions

BACKGROUND

Emergency medical service (EMS) data, particularly from the emergency department (ED), is a common source of information for syndromic surveillance. However, the entire EMS chain, consists of both out-of-hospital and in-hospital services. Differences in validity and timeliness across these data sources so far have not been studied. Neither have the differences in validity and timeliness of this data from different European countries. In this paper we examine the validity and timeliness of the entire chain of EMS data sources from three European regions for common syndromic influenza surveillance during the A(H1N1) influenza pandemic in 2009.

METHODS

We gathered local, regional, or national information on influenza-like illness (ILI) or respiratory syndrome from an Austrian Emergency Medical Dispatch Service (EMD-AT), an Austrian and Belgian ambulance services (EP-AT, EP-BE) and from a Belgian and Spanish emergency department (ED-BE, ED-ES). We examined the timeliness of the EMS data in identifying the beginning of the autumn/winter wave of pandemic A(H1N1) influenza as compared to the reference data. Additionally, we determined the sensitivity and specificity of an aberration detection algorithm (Poisson CUSUM) in EMS data sources for detecting the autumn/winter wave of the A(H1N1) influenza pandemic.

RESULTS

The ED-ES data demonstrated the most favourable validity, followed by the ED-BE data. The beginning of the autumn/winter wave of pandemic A(H1N1) influenza was identified eight days in advance in ED-BE data. The EP data performed stronger in data sets for large catchment areas (EP-BE) and identified the beginning of the autumn/winter wave almost at the same time as the reference data (time lag +2 days). EMD data exhibited timely identification of the autumn/winter wave of A(H1N1) but demonstrated weak validity measures.

CONCLUSIONS

In this study ED data exhibited the most favourable performance in terms of validity and timeliness for syndromic influenza surveillance, along with EP data for large catchment areas. For the other data sources performance assessment delivered no clear results. The study shows that routinely collected data from EMS providers can augment and enhance public health surveillance of influenza by providing information during health crises in which such information must be both timely and readily obtainable.

Finding evidence for local transmission of contagious disease in molecular epidemiological datasets

Surveillance systems of contagious diseases record information on cases to monitor incidence of disease and to evaluate effectiveness of interventions. These systems focus on a well-defined population; a key question is whether observed cases are infected through local transmission within the population or whether cases are the result of importation of infection into the population. Local spread of infection calls for different intervention measures than importation of infection. Besides standardized information on time of symptom onset and location of cases, pathogen genotyping or sequencing offers essential information to address this question. Here we introduce a method that takes full advantage of both the genetic and epidemiological data to distinguish local transmission from importation of infection, by comparing inter-case distances in temporal, spatial and genetic data. Cases that are part of a local transmission chain will have shorter distances between their geographical locations, shorter durations between their times of symptom onset and shorter genetic distances between their pathogen sequences as compared to cases that are due to importation. In contrast to generic clustering algorithms, the proposed method explicitly accounts for the fact that during local transmission of a contagious disease the cases are caused by other cases. No pathogen-specific assumptions are needed due to the use of ordinal distances, which allow for direct comparison between the disparate data types. Using simulations, we test the performance of the method in identifying local transmission of disease in large datasets, and assess how sensitivity and specificity change with varying size of local transmission chains and varying overall disease incidence.

Syndromic surveillance using veterinary laboratory data: data pre-processing and algorithm performance evaluation

Diagnostic test orders to an animal laboratory were explored as a data source for monitoring trends in the incidence of clinical syndromes in cattle. Four years of real data and over 200 simulated outbreak signals were used to compare pre-processing methods that could remove temporal effects in the data, as well as temporal aberration detection algorithms that provided high sensitivity and specificity. Weekly differencing demonstrated solid performance in removing day-of-week effects, even in series with low daily counts. For aberration detection, the results indicated that no single algorithm showed performance superior to all others across the range of outbreak scenarios simulated. Exponentially weighted moving average charts and Holt-Winters exponential smoothing demonstrated complementary performance, with the latter offering an automated method to adjust to changes in the time series that will likely occur in the future. Shewhart charts provided lower sensitivity but earlier detection in some scenarios. Cumulative sum charts did not appear to add value to the system; however, the poor performance of this algorithm was attributed to characteristics of the data monitored. These findings indicate that automated monitoring aimed at early detection of temporal aberrations will likely be most effective when a range of algorithms are implemented in parallel.

CUSUM Residual Charts for Monitoring Enterovirus Infections

We consider the syndromic surveillance problem for enterovirus (EV) like cases. The data used in this study are the daily counts of EV-like cases sampled from the National Health Insurance Research Database in Taiwan. To apply the CUSUM procedure for syndromic surveillance, a regression model with time-series error-term is used. Our results show that the CUSUM chart is helpful to detect abnormal increases of the visit frequency.

Retrospective time series analysis of veterinary laboratory data: preparing a historical baseline for cluster detection in syndromic surveillance

The practice of disease surveillance has shifted in the last two decades towards the introduction of systems capable of early detection of disease. Modern biosurveillance systems explore different sources of pre-diagnostic data, such as patient's chief complaint upon emergency visit or laboratory test orders. These sources of data can provide more rapid detection than traditional surveillance based on case confirmation, but are less specific, and therefore their use poses challenges related to the presence of background noise and unlabelled temporal aberrations in historical data. The overall goal of this study was to carry out retrospective analysis using three years of laboratory test submissions to the Animal Health Laboratory in the province of Ontario, Canada, in order to prepare the data for use in syndromic surveillance. Daily cases were grouped into syndromes and counts for each syndrome were monitored on a daily basis when medians were higher than one case per day, and weekly otherwise. Poisson regression accounting for day-of-week and month was able to capture the day-of-week effect with minimal influence from temporal aberrations. Applying Poisson regression in an iterative manner, that removed data points above the predicted 95th percentile of daily counts, allowed for the removal of these aberrations in the absence of labelled outbreaks, while maintaining the day-of-week effect that was present in the original data. This resulted in the construction of time series that represent the baseline patterns over the past three years, free of temporal aberrations. The final method was thus able to remove temporal aberrations while keeping the original explainable effects in the data, did not need a training period free of aberrations, had minimal adjustment to the aberrations present in the raw data, and did not require labelled outbreaks. Moreover, it was readily applicable to the weekly data by substituting Poisson regression with moving 95th percentiles.

FluBreaks: early epidemic detection from Google flu trends

BACKGROUND

The Google Flu Trends service was launched in 2008 to track changes in the volume of online search queries related to flu-like symptoms. Over the last few years, the trend data produced by this service has shown a consistent relationship with the actual number of flu reports collected by the US Centers for Disease Control and Prevention (CDC), often identifying increases in flu cases weeks in advance of CDC records. However, contrary to popular belief, Google Flu Trends is not an early epidemic detection system. Instead, it is designed as a baseline indicator of the trend, or changes, in the number of disease cases.

OBJECTIVE

To evaluate whether these trends can be used as a basis for an early warning system for epidemics.

METHODS

We present the first detailed algorithmic analysis of how Google Flu Trends can be used as a basis for building a fully automated system for early warning of epidemics in advance of methods used by the CDC. Based on our work, we present a novel early epidemic detection system, called FluBreaks (dritte.org/flubreaks), based on Google Flu Trends data. We compared the accuracy and practicality of three types of algorithms: normal distribution algorithms, Poisson distribution algorithms, and negative binomial distribution algorithms. We explored the relative merits of these methods, and related our findings to changes in Internet penetration and population size for the regions in Google Flu Trends providing data.

RESULTS

Across our performance metrics of percentage true-positives (RTP), percentage false-positives (RFP), percentage overlap (OT), and percentage early alarms (EA), Poisson- and negative binomial-based algorithms performed better in all except RFP. Poisson-based algorithms had average values of 99%, 28%, 71%, and 76% for RTP, RFP, OT, and EA, respectively, whereas negative binomial-based algorithms had average values of 97.8%, 17.8%, 60%, and 55% for RTP, RFP, OT, and EA, respectively. Moreover, the EA was also affected by the region's population size. Regions with larger populations (regions 4 and 6) had higher values of EA than region 10 (which had the smallest population) for negative binomial- and Poisson-based algorithms. The difference was 12.5% and 13.5% on average in negative binomial- and Poisson-based algorithms, respectively.

CONCLUSIONS

We present the first detailed comparative analysis of popular early epidemic detection algorithms on Google Flu Trends data. We note that realizing this opportunity requires moving beyond the cumulative sum and historical limits method-based normal distribution approaches, traditionally employed by the CDC, to negative binomial- and Poisson-based algorithms to deal with potentially noisy search query data from regions with varying population and Internet penetrations. Based on our work, we have developed FluBreaks, an early warning system for flu epidemics using Google Flu Trends.

Surveillance of community outbreaks of respiratory tract infections based on house-call visits in the metropolitan area of Athens, Greece

BACKGROUND

The traditional Serfling-type approach for influenza-like illness surveillance requires long historical time-series. We retrospectively evaluated the use of recent, short, historical time-series for recognizing the onset of community outbreaks of respiratory tract infections (RTIs).

METHODS

The data used referred to the proportion of diagnoses for upper or lower RTIs to total diagnoses for house-call visits, performed by a private network of medical specialists (SOS Doctors) in the metropolitan area of Athens, Greece, between January 01, 2000 and October 12, 2008. The reference standard classification of the observations was obtained by generating epidemic thresholds after analyzing the full 9-year period. We evaluated two different alert generating methods [simple regression and cumulative sum (CUSUM), respectively], under a range of input parameters, using data for the previous running 4-6 week period. These methods were applied if the previous weeks contained non-aberrant observations.

RESULTS

We found that the CUSUM model with a specific set of parameters performed marginally better than simple regression for both groups. The best results (sensitivity, specificity) for simple regression and CUSUM models for upper RTIs were (1.00, 0.82) and (0.94, 0.93) respectively. Corresponding results for lower RTIs were (1.00, 0.80) and (0.93, 0.91) respectively.

CONCLUSIONS

Short-term data for house-call visits can be used rather reliably to identify respiratory tract outbreaks in the community using simple regression and CUSUM methods. Such surveillance models could be particularly useful when a large historical database is either unavailable or inaccurate and, thus, traditional methods are not optimal.

Incidence analyses and space-time cluster detection of hepatitis C in Fujian Province of China from 2006 to 2010

BACKGROUND

There is limited epidemiologic information about the incidence of hepatitis C in China, and few studies have applied space-time scan statistic to detect clusters of hepatitis C and made adjustment for temporal trend and relative risk of regions.

METHODOLOGY AND PRINCIPAL FINDINGS

We analyzed the temporal changes and characteristics of incidence of hepatitis C in Fujian Province from 2006 through 2010. The discrete Poisson model of space-time scan statistic was chosen for cluster detection. Data on new cases of hepatitis C were obtained from the Center for Disease Control and Prevention of Fujian Province. Between 2006 and 2010, there was an annualized increase in the incidence of hepatitis C of 23.0 percent, from 928 cases (2.63 per 100,000 persons) to 2,180 cases (6.01 per 100,000 persons). The incidence among women increased more rapidly. The cumulative incidence showed that people who were over 60 years had the highest risk to suffer hepatitis C (52.51 per 100,000 persons), and women had lower risk compared to men (OR=0.69). Putian had the highest cumulative incidence among all the regions (86.95 per 100,000 persons). The most likely cluster was identified in Putian during March to August in 2009 without adjustment, but it shifted to three contiguous cities with a two-month duration after adjustment for temporal trend and relative risk of regions.

CONCLUSIONS/SIGNIFICANCE

The incidence of hepatitis C is increasing in Fujian Province, and women are at a more rapid pace. The space-time scan statistic is useful as a screening tool for clusters of hepatitis C, with adjustment for temporal trend and relative risk of regions recommended.

Epidemic features affecting the performance of outbreak detection algorithms

BACKGROUND

Outbreak detection algorithms play an important role in effective automated surveillance. Although many algorithms have been designed to improve the performance of outbreak detection, few published studies have examined how epidemic features of infectious disease impact on the detection performance of algorithms. This study compared the performance of three outbreak detection algorithms stratified by epidemic features of infectious disease and examined the relationship between epidemic features and performance of outbreak detection algorithms.

METHODS

Exponentially weighted moving average (EWMA), cumulative sum (CUSUM) and moving percentile method (MPM) algorithms were applied. We inserted simulated outbreaks into notifiable infectious disease data in China Infectious Disease Automated-alert and Response System (CIDARS), and compared the performance of the three algorithms with optimized parameters at a fixed false alarm rate of 5% classified by epidemic features of infectious disease. Multiple linear regression was adopted to analyse the relationship of the algorithms' sensitivity and timeliness with the epidemic features of infectious diseases.

RESULTS

The MPM had better detection performance than EWMA and CUSUM through all simulated outbreaks, with or without stratification by epidemic features (incubation period, baseline counts and outbreak magnitude). The epidemic features were associated with both sensitivity and timeliness. Compared with long incubation, short incubation had lower probability (β* = -0.13, P < 0.001) but needed shorter time to detect outbreaks (β* = -0.57, P < 0.001). Lower baseline counts were associated with higher probability (β* = -0.20, P < 0.001) and longer time (β* = 0.14, P < 0.001). The larger outbreak magnitude was correlated with higher probability (β* = 0.55, P < 0.001) and shorter time (β* = -0.23, P < 0.001).

CONCLUSIONS

The results of this study suggest that the MPM is a prior algorithm for outbreak detection and differences of epidemic features in detection performance should be considered in automatic surveillance practice.

Probabilistic case detection for disease surveillance using data in electronic medical records

This paper describes a probabilistic case detection system (CDS) that uses a Bayesian network model of medical diagnosis and natural language processing to compute the posterior probability of influenza and influenza-like illness from emergency department dictated notes and laboratory results. The diagnostic accuracy of CDS for these conditions, as measured by the area under the ROC curve, was 0.97, and the overall accuracy for NLP employed in CDS was 0.91.

Detecting disease outbreaks using local spatiotemporal methods

A real-time surveillance method is developed with emphasis on rapid and accurate detection of emerging outbreaks. We develop a model with relatively weak assumptions regarding the latent processes generating the observed data, ensuring a robust prediction of the spatiotemporal incidence surface. Estimation occurs via a local linear fitting combined with day-of-week effects, where spatial smoothing is handled by a novel distance metric that adjusts for population density. Detection of emerging outbreaks is carried out via residual analysis. Both daily residuals and AR model-based detrended residuals are used for detecting abnormalities in the data given that either a large daily residual or an increasing temporal trend in the residuals signals a potential outbreak, with the threshold for statistical significance determined using a resampling approach.

Adjusting outbreak detection algorithms for surveillance during epidemic and non-epidemic periods

Many aberration detection algorithms are used in infectious disease surveillance systems to assist in the early detection of potential outbreaks. In this study, we explored a novel approach to adjusting aberration detection algorithms to account for the impact of seasonality inherent in some surveillance data. By using surveillance data for hand-foot-and-mouth disease in Shandong province, China, we evaluated the use of seasonally-adjusted alerting thresholds with three aberration detection methods (C1, C2, and C3). We found that the optimal thresholds of C1, C2, and C3 varied between the epidemic and non-epidemic seasons of hand-foot-and-mouth disease, and the application of seasonally adjusted thresholds improved the performance of outbreak detection by maintaining the same sensitivity and timeliness while decreasing by nearly half the false alert rate during the non-epidemic season. Our preliminary findings suggest a general approach to improving aberration detection for outbreaks of infectious disease with seasonally variable incidence.

[Integrated bulletin for the automated surveillance of notifiable communicable diseases in Schleswig-Holstein (IBISSH). A spatiotemporal early warning system]

The program package "Integrated Bulletin for Infectious disease Surveillance for Schleswig-Holstein" (IBIS(SH)) was introduced in 2008 for the automated data analysis of notifiable infectious diseases in Schleswig-Holstein, Germany. The Java-based IBIS(SH) software supports access to the national SurvNet@RKI reporting data via Access and MS SQL. The aim of the IBIS(SH) system is early warning and interpretation of clusters and monitoring of trends. One module of the system permits the analysis of temporal aberration by comparison of data from previous years. The interpretation system is based on the weekly median and on percentile values from previous years. The extent of an aberration is assessed by a five-step score magnitude scale. Another module permits the detection of regional clusters by the weekly assessment of a population-based risk analysis. IBIS(SH) automatically generates tables and graphs for the weekly bulletin and their allocation in the Internet. Data for the most relevant pathogens in Schleswig-Holstein are presented for the year 2009. The performance of the automated temporal and the regional detection systems are compared to outbreak detection by local health authorities.

Rapid genoserotyping tool for classification of Salmonella serovars

We have developed a Salmonella genoserotyping array (SGSA) which rapidly generates an antigenic formula consistent with the White-Kauffmann-Le Minor scheme, currently the gold standard for Salmonella serotyping. A set of 287 strains representative of 133 Salmonella serovars was assembled to validate the array and to test the array probes for accuracy, specificity, and reproducibility. Initially, 76 known serovars were utilized to validate the specificity and repeatability of the array probes and their expected probe patterns. The SGSA generated the correct serovar designations for 100% of the known subspecies I serovars tested in the validation panel and an antigenic formula consistent with that of the White-Kauffmann-Le Minor scheme for 97% of all known serovars tested. Once validated, the SGSA was assessed against a blind panel of 100 Salmonella enterica subsp. I samples serotyped using traditional methods. In summary, the SGSA correctly identified all of the blind samples as representing Salmonella and successfully identified 92% of the antigens found within the unknown samples. Antigen- and serovar-specific probes, in combination with a pepT PCR for confirmation of S. enterica subsp. Enteritidis determinations, generated an antigenic formula and/or a serovar designation consistent with the White-Kauffmann-Le Minor scheme for 87% of unknown samples tested with the SGSA. Future experiments are planned to test the specificity of the array probes with other Salmonella serovars to demonstrate the versatility and utility of this array as a public health tool in the identification of Salmonella.

Hand, foot and mouth disease: spatiotemporal transmission and climate

BACKGROUND

The Hand-Foot-Mouth Disease (HFMD) is the most common infectious disease in China, its total incidence being around 500,000~1,000,000 cases per year. The composite space-time disease variation is the result of underlining attribute mechanisms that could provide clues about the physiologic and demographic determinants of disease transmission and also guide the appropriate allocation of medical resources to control the disease.

METHODS AND FINDINGS

HFMD cases were aggregated into 1456 counties and during a period of 11 months. Suspected climate attributes to HFMD were recorded monthly at 674 stations throughout the country and subsequently interpolated within 1456 × 11 cells across space-time (same as the number of HFMD cases) using the Bayesian Maximum Entropy (BME) method while taking into consideration the relevant uncertainty sources. The dimensionalities of the two datasets together with the integrated dataset combining the two previous ones are very high when the topologies of the space-time relationships between cells are taken into account. Using a self-organizing map (SOM) algorithm the dataset dimensionality was effectively reduced into 2 dimensions, while the spatiotemporal attribute structure was maintained. 16 types of spatiotemporal HFMD transmission were identified, and 3-4 high spatial incidence clusters of the HFMD types were found throughout China, which are basically within the scope of the monthly climate (precipitation) types.

CONCLUSIONS

HFMD propagates in a composite space-time domain rather than showing a purely spatial and purely temporal variation. There is a clear relationship between HFMD occurrence and climate. HFMD cases are geographically clustered and closely linked to the monthly precipitation types of the region. The occurrence of the former depends on the later.

Development of a Salmonella screening tool for consumer complaint-based foodborne illness surveillance systems

Foodborne illness surveillance based on consumer complaints detects outbreaks by finding common exposures among callers, but this process is often difficult. Laboratory testing of ill callers could also help identify potential outbreaks. However, collection of stool samples from all callers is not feasible. Methods to help screen calls for etiology are needed to increase the efficiency of complaint surveillance systems and increase the likelihood of detecting foodborne outbreaks caused by Salmonella. Data from the Minnesota Department of Health foodborne illness surveillance database (2000 to 2008) were analyzed. Complaints with identified etiologies were examined to create a predictive model for Salmonella. Bootstrap methods were used to internally validate the model. Seventy-one percent of complaints in the foodborne illness database with known etiologies were due to norovirus. The predictive model had a good discriminatory ability to identify Salmonella calls. Three cutoffs for the predictive model were tested: one that maximized sensitivity, one that maximized specificity, and one that maximized predictive ability, providing sensitivities and specificities of 32 and 96%, 100 and 54%, and 89 and 72%, respectively. Development of a predictive model for Salmonella could help screen calls for etiology. The cutoff that provided the best predictive ability for Salmonella corresponded to a caller reporting diarrhea and fever with no vomiting, and five or fewer people ill. Screening calls for etiology would help identify complaints for further follow-up and result in identifying Salmonella cases that would otherwise go unconfirmed; in turn, this could lead to the identification of more outbreaks.

Rapid detection of pandemic influenza in the presence of seasonal influenza

BACKGROUND

Key to the control of pandemic influenza are surveillance systems that raise alarms rapidly and sensitively. In addition, they must minimise false alarms during a normal influenza season. We develop a method that uses historical syndromic influenza data from the existing surveillance system 'SERVIS' (Scottish Enhanced Respiratory Virus Infection Surveillance) for influenza-like illness (ILI) in Scotland.

METHODS

We develop an algorithm based on the weekly case ratio (WCR) of reported ILI cases to generate an alarm for pandemic influenza. From the seasonal influenza data from 13 Scottish health boards, we estimate the joint probability distribution of the country-level WCR and the number of health boards showing synchronous increases in reported influenza cases over the previous week. Pandemic cases are sampled with various case reporting rates from simulated pandemic influenza infections and overlaid with seasonal SERVIS data from 2001 to 2007. Using this combined time series we test our method for speed of detection, sensitivity and specificity. Also, the 2008-09 SERVIS ILI cases are used for testing detection performances of the three methods with a real pandemic data.

RESULTS

We compare our method, based on our simulation study, to the moving-average Cumulative Sums (Mov-Avg Cusum) and ILI rate threshold methods and find it to be more sensitive and rapid. For 1% case reporting and detection specificity of 95%, our method is 100% sensitive and has median detection time (MDT) of 4 weeks while the Mov-Avg Cusum and ILI rate threshold methods are, respectively, 97% and 100% sensitive with MDT of 5 weeks. At 99% specificity, our method remains 100% sensitive with MDT of 5 weeks. Although the threshold method maintains its sensitivity of 100% with MDT of 5 weeks, sensitivity of Mov-Avg Cusum declines to 92% with increased MDT of 6 weeks. For a two-fold decrease in the case reporting rate (0.5%) and 99% specificity, the WCR and threshold methods, respectively, have MDT of 5 and 6 weeks with both having sensitivity close to 100% while the Mov-Avg Cusum method can only manage sensitivity of 77% with MDT of 6 weeks. However, the WCR and Mov-Avg Cusum methods outperform the ILI threshold method by 1 week in retrospective detection of the 2009 pandemic in Scotland.

CONCLUSIONS

While computationally and statistically simple to implement, the WCR algorithm is capable of raising alarms, rapidly and sensitively, for influenza pandemics against a background of seasonal influenza. Although the algorithm was developed using the SERVIS data, it has the capacity to be used at other geographic scales and for different disease systems where buying some early extra time is critical.

Detecting new and emerging diseases on livestock farms using an early detection system

The monitoring and surveillance of animal diseases is becoming increasingly important to policy-makers in Great Britain particularly given recent incursions of avian influenza and the emergence of bovine spongiform encephalopathy. To meet this surveillance objective, data from British livestock is collected and analysed retrospectively on an ongoing basis. However, these data can also be analysed prospectively within an early detection system which raises alerts to significant increases in disease reporting soon after they occur in the field. The feasibility of such an approach has been examined previously for Salmonella. This paper applied the approach to a further subset of surveillance data to alert those monitoring disease to increases in potentially new and emerging diseases. Thus far, the analysis, conducted on a quarterly basis, has proved a useful additional tool in enhanced surveillance by raising alerts to significant increases in several syndromes in both sheep and cattle.