Case Detection

Influenza causes significant morbidity and mortality, with attendant costs of roughly $10 billion for treatment and up to $77 billion in indirect costs annually. The Centers for Disease Control and Prevention conducts annual influenza surveillance, and includes measures of inpatient and outpatient influenza-related activity, disease severity, and geographic spread. However, inherent lags in the current methods used for data collection and transmission result in a one to two weeks delay in notification of an outbreak via the Centers for Disease Control and Prevention’s website. Early notification might facilitate clinical decision-making when patients present with acute respiratory infection during the early stages of the influenza outbreak. 

In the United States, the influenza surveillance season typically begins in October and continues through May. The Utah Health Research Network has participated in Centers for Disease Control and Prevention’s influenza surveillance since 2002, collecting data on outpatient visits for influenza-like illness (ILI, defined as fever of 100F or higher with either cough or sore throat). Over time, Utah Health Research Network has moved from data collection by hand to automated data collection that extracts information from discrete fields in patients’ electronic health records.

We used statistical process control to generate surveillance graphs of ILI and positive rapid influenza tests, using data available electronically from the electronic health records. 

Objective

The objective of this study is to describe the use of point-of-care rapid influenza testing in an outpatient, setting for the identification of the onset of influenza in the community. 

Group A Streptococcal (GAS) pharyngitis, the most common bacterial cause of acute pharyngitis, causes more than half a billion cases annually worldwide. Treatment with antibiotics provides symptomatic benefit and reduces complications, missed work days and transmission. Physical examination alone is an unreliable way to distinguish GAS from other causes of pharyngitis, so the 4-point Centor score, based on history and physical, is used to classify GAS risk. Still, patients with pharyngitis are often misclassified, leading to inappropriate antibiotic treatment of those with viral disease and to under-treatment of those with bone fide GAS. One key problem, even when clinical guidelines are followed, is that diagnostic accuracy for GAS pharyngitis is affected by earlier probability of disease, which in turn is related to exposure. Point-of-care clinicians rarely have access to valuable biosurveillance-derived contextualizing information when making clinical management decisions.

Objective

The objective of this study was to measure the value of integrating real-time contemporaneous local disease incidence (biosurveillance) data with a clinical score, to more accurately identify patients with GAS pharyngitis.

The EPA Water Security initiative contamination warning system detection strategy involves the use of multiple monitoring and surveillance components for timely detection of drinking water contamination in the distribution system. The public health surveillance (PHS) component of the contamination warning system involves the analysis of health-related data to identify disease events that may stem from drinking water contamination. Public health data include hospital admission reports, infectious disease surveillance, emergency medical service reports, 911 calls and poison control center calls. Automated analysis of these data streams results in alerts, which are investigated by health department epidemiologists. A comprehensive operational strategy was developed to describe the processes and procedures involved in the the initial investigation and validation of a PHS alert. The operational strategy established specific roles and responsibilities, and detailed procedural flow descriptions. The procedural flow concluded with the determination of whether or not an alert generated from surveillance of public health data streams is indicative of a possible water contamination incident.

Objective

To develop standard operating procedures to identify or rule out possible water contamination as a cause for a syndromic surveillance alarm.

The summer of 2010 in Maryland was characterized by unusually high temperatures. This type of increased and prolonged heat can potentially make residents sick, and extreme exposure can even kill people at highest risk. Numerous deaths throughout the state were attributed to this heat wave. The Maryland Department of Health and Mental Hygiene addressed this public health issue by using public messaging and maintaining constant situational awareness through the electronic syndromic surveillance. Thus, the electronic surveillance system for the early notification of community-based epidemics (ESSENCE) was used to monitor heat-related illnesses throughout the state.

Objective

This paper describes the use of ESSENCE, a syndromic surveillance system, to monitor heat-related illnesses throughout the state of Maryland during the summer of 2010.

On 24 December 2009, a female New Hampshire resident was confirmed to have gastrointestinal anthrax on the basis of clinical findings and laboratory testing. Her source of anthrax was not immediately known, so the New Hampshire Department of Health and Human Services, in conjunction with several other state and federal agencies, conducted a comprehensive epidemiologic investigation, which included active surveillance to identify any additional anthrax cases from a similar exposure. It was determined that the index patient participated in a drumming event with animal-hide drums on 4 December, one day before the onset of symptoms. Two drums used at the event were later found to be contaminated with Bacillus anthracis.

Objective

This paper describes the use of customizable tools to query electronic emergency department data, as part of case finding, during the response to a community anthrax exposure in New Hampshire.

Current practices of automated case detection fall into the extremes of diagnostic accuracy and timeliness. In regards to diagnostic accuracy, electronic laboratory reporting (ELR) is at one extreme and syndromic surveillance is at the other. In regards to timeliness, syndromic surveillance can be immediate, and ELR is delayed 7 days from initial patient visit. A plausible solution, a middle way, to the extremes of diagnostic precision and timeliness in current case detection practices is an automated Bayesian diagnostic system that uses all available data types, for example, freetext ED reports, radiology reports, and laboratory reports.We have built such a solution - Bayesian case detection (BCD). As a probabilistic system, BCD operates across the spectrum of diagnostic accuracy, that is, it outputs the degree of certainty for every diagnosis. In addition, BCD incorporates multiple data types as they appear during the course of a patient encounter or lifetime, with no degradation in the ability to perform diagnosis.

Objective

This paper describes the architecture and evaluation of our recently developed automated BCD system.

Event-based biosurveillance is a practice of monitoring diverse information sources for the detection of events pertaining to human health. Online documents, such as news articles on the Internet, have commonly been the primary information sources in event-based biosurveillance. With the large number of online publications as well as with the language diversity, thorough monitoring of online documents is challenging. Automated document classification is an important step toward efficient event-based biosurveillance. In Project Argus, a biosurveillance program hosted at Georgetown University Medical Center, supervised and unsupervised approaches to document classification are considered for event-based biosurveillance.

Objective

This paper describes ongoing efforts in enhancing automated document classification toward efficient event-based biosurveillance. 

Scan statistics are highly successful for the evaluation of space-time clusters. Recently, concepts from the graph theory were applied to evaluate the set of potential clusters. Wieland et al. introduced a graph theoretical method for detecting arbitrarily shaped clusters on the basis of the Euclidean minimum spanning tree of cartogram transformed case locations, which is quite effective, but the cartogram construction step of this algorithm is computationally expensive and complicated.

Objective

We describe a method for prospective space-time cluster detection of point event data based on the scan statistic. Our aim is to detect as early as possible the appearance of an emerging cluster of syndromic individuals because of a real outbreak of disease amidst the heterogeneous population at risk.

The research reported in this paper is part of a larger effort to achieve better signal-to-noise ratio, hence accuracy, in pharmacovigilance applications. The relatively low frequency of occurrence of adverse drug reactions leads to weak causal relations between the reaction and any measured signal. We hypothesize that by grouping related signals, we can enhance detection rate and suppress false alarm rate.

Objective

ICD-9 codes are commonly used to identify disease cohorts and are often found to be less than adequate. Data available in structured databasesFlab test results, medications etc.Fcan supplement the diagnosis codes. In this study, we describe an automated method that uses these related data items, and no additional manual annotations to more accurately identify patient cohorts.