William R. Hogan

Accuracy of Data in Computer-based Patient Records

Data in computer-based patient records (CPRs) have many uses beyond their primary role in patient care, including research and health-system management. Although the accuracy of CPR data directly affects these applications, there has been only sporadic interest in, and no previous review of, data accuracy in CPRs. This paper reviews the published studies of data accuracy in CPRs. These studies report highly variable levels of accuracy. This variability stems from differences in study design, in types of data studied, and in the CPRs themselves. These differences confound interpretation of this literature. We conclude that our knowledge of data accuracy in CPRs is not commensurate with its importance and further studies are needed. We propose methodological guidelines for studying accuracy that address shortcomings of the current literature. As CPR data are used increasingly for research, methods used in research databases to continuously monitor and improve accuracy should be applied to CPRs.

Algorithms for rapid outbreak detection: a research synthesis

The threat of bioterrorism has stimulated interest in enhancing public health surveillance to detect disease outbreaks more rapidly than is currently possible. To advance research on improving the timeliness of outbreak detection, the Defense Advanced Research Project Agency sponsored the Bio-event Advanced Leading Indicator Recognition Technology (BioALIRT) project beginning in 2001. The purpose of this paper is to provide a synthesis of research on outbreak detection algorithms conducted by academic and industrial partners in the BioALIRT project. We first suggest a practical classification for outbreak detection algorithms that considers the types of information encountered in surveillance analysis. We then present a synthesis of our research according to this classification. The research conducted for this project has examined how to use spatial and other covariate information from disparate sources to improve the timeliness of outbreak detection. Our results suggest that use of spatial and other covariate information can improve outbreak detection performance. We also identified, however, methodological challenges that limited our ability to determine the benefit of using outbreak detection algorithms that operate on large volumes of data. Future research must address challenges such as forecasting expected values in high-dimensional data and generating spatial and multivariate test data sets.

The Accuracy of Medication Data in an Outpatient Electronic Medical Record

Objective: To measure the accuracy-of medication records stored in the electronic medical record (EMR) of an outpatient geriatric center. The authors analyzed accuracy from the perspective of a clinician using the data and the perspective of a computer-based medical decision-support system (MDSS). Design: Prospective cohort study. Methods: The EMR at the geriatric center captures medication data both directly from clinicians and indirectly using encounter forms and data-entry clerks. During a scheduled office visit for medical care, the treating clinician determined whether the medication records for the patient were an accurate representation of the medications that the patient was actually taking. Using the available sources of information (the patient, the patient’s vials, any caregivers, and the medical chart), the clinician determined whether the recorded data were correct, whether any data were missing, and the type and cause for each discrepancy found. Results: At the geriatric center, 83% of medication records represented correctly the compound, dose, and schedule of a current medication; 91% represented correctly the compound. 0.37 current medications were missing per patient. The principal cause of errors was the patient (36.1% of errors), who misreported a medication at a previous visit or changed (stopped, started, or dose-adjusted) a medication between visits. The second most frequent cause of errors was failure to capture changes to medications made by outside clinicians, accounting for 25.9% of errors. Transcription errors were a relatively ucommon cause (8.2% of errors). When the accuracy of records from the center was analyzed from the perspective of a MDSS, 90% were correct for compound identity and 1.38 medications were missing or uncoded per patient. The cause of the additional errors of omission was a free-text “comments” field-which it is assumed would be unreadable by current MDSS applications-that was used by clinicians in 18% of records to record the identity of the medication. Conclusions: Medication records in an outpatient EMR may have significant levels of data error. Based on an analysis of correctable causes of error, the authors conclude that the most effective extension to the EMR studied would be to expand its scope to include all clinicians who can potentially change medications. Even with EMR extensions, however, ineradicable error due to patients and data entry will remain. Several implications of ineradicable error for MDSSs are discussed. The provision of a free-text “comments” field increased the accuracy of medication lists for clinician users at the expense of accuracy for a MDSS.

Methodological Review: Natural Language Processing methods and systems for biomedical ontology learning

While the biomedical informatics community widely acknowledges the utility of domain ontologies, there remain many barriers to their effective use. One important requirement of domain ontologies is that they must achieve a high degree of coverage of the domain concepts and concept relationships. However, the development of these ontologies is typically a manual, time-consuming, and often error-prone process. Limited resources result in missing concepts and relationships as well as difficulty in updating the ontology as knowledge changes. Methodologies developed in the fields of Natural Language Processing, information extraction, information retrieval and machine learning provide techniques for automating the enrichment of an ontology from free-text documents. In this article, we review existing methodologies and developed systems, and discuss how existing methods can benefit the development of biomedical ontologies.

Design of a national retail data monitor for public health surveillance.

The National Retail Data Monitor receives data daily from 10,000 stores, including pharmacies, that sell health care products. These stores belong to national chains that process sales data centrally and utilize Universal Product Codes and scanners to collect sales information at the cash register. The high degree of retail sales data automation enables the monitor to collect information from thousands of store locations in near to real time for use in public health surveillance. The monitor provides user interfaces that display summary sales data on timelines and maps. Algorithms monitor the data automatically on a daily basis to detect unusual patterns of sales. The project provides the resulting data and analyses, free of charge, to health departments nationwide. Future plans include continued enrollment and support of health departments, developing methods to make the service financially self-supporting, and further refinement of the data collection system to reduce the time latency of data receipt and analysis.

EPO-KB

UNLABELLED The knowledge base EPO-KB (Empirical Proteomic Ontology Knowledge Base) is based on an OWL ontology that represents current knowledge linking mass-to-charge (m/z) ratios to proteins on multiple platforms including Matrix Assisted Laser/Desorption Ionization (MALDI) and Surface Enhanced Laser/Desorption Ionization (SELDI)--Time of Flight (TOF). At present, it contains information on m/z ratio to protein links that were extracted from 120 published research papers. It has a web interface that allows researchers to query and retrieve putative proteins that correspond to a user-specified m/z ratio. EPO-KB also allows automated entry of additional m/z ratio to protein links and is expandable to the addition of gene to protein and protein to disease links. AVAILABILITY http://www.dbmi.pitt.edu/EPO-KB

Developing a semantically rich ontology for the biobank-administration domain

BackgroundBiobanks are a critical resource for translational science. Recently, semantic web technologies such as ontologies have been found useful in retrieving research data from biobanks. However, recent research has also shown that there is a lack of data about the administrative aspects of biobanks. These data would be helpful to answer research-relevant questions such as what is the scope of specimens collected in a biobank, what is the curation status of the specimens, and what is the contact information for curators of biobanks. Our use cases include giving researchers the ability to retrieve key administrative data (e.g. contact information, contacts affiliation, etc.) about the biobanks where specific specimens of interest are stored. Thus, our goal is to provide an ontology that represents the administrative entities in biobanking and their relations. We base our ontology development on a set of 53 data attributes called MIABIS, which were in part the result of semantic integration efforts of the European Biobanking and Biomolecular Resources Research Infrastructure (BBMRI). The previous work on MIABIS provided the domain analysis for our ontology. We report on a test of our ontology against competency questions that we derived from the initial BBMRI use cases. Future work includes additional ontology development to answer additional competency questions from these use cases.ResultsWe created an open-source ontology of biobank administration called Ontologized MIABIS (OMIABIS) coded in OWL 2.0 and developed according to the principles of the OBO Foundry. It re-uses pre-existing ontologies when possible in cooperation with developers of other ontologies in related domains, such as the Ontology of Biomedical Investigation. OMIABIS provides a formalized representation of biobanks and their administration. Using the ontology and a set of Description Logic queries derived from the competency questions that we identified, we were able to retrieve test data with perfect accuracy. In addition, we began development of a mapping from the ontology to pre-existing biobank data structures commonly used in the U.S.ConclusionsIn conclusion, we created OMIABIS, an ontology of biobank administration. We found that basing its development on pre-existing resources to meet the BBMRI use cases resulted in a biobanking ontology that is re-useable in environments other than BBMRI. Our ontology retrieved all true positives and no false positives when queried according to the competency questions we derived from the BBMRI use cases. Mapping OMIABIS to a data structure used for biospecimen collections in a medical center in Little Rock, AR showed adequate coverage of our ontology.

Unsupervised clustering of over-the-counter healthcare products into product categories

A general problem in biosurveillance is finding appropriate aggregates of elemental data to monitor for the detection of disease outbreaks. We developed an unsupervised clustering algorithm for aggregating over-the-counter healthcare (OTC) products into categories. This algorithm employs MCMC over hundreds of parameters in a Bayesian model to place products into clusters. Despite the high dimensionality, it still performs fast on hundreds of time series. The procedure was able to uncover a clinically significant distinction between OTC products intended for the treatment of allergy and OTC products intended for the treatment of cough, cold, and influenza symptoms.

Prescription-Acquired Acetaminophen Use and the Risk of Asthma in Adults: A Case-Control Study

Background: Studies have examined the association between acetaminophen use and asthma; however, their interpretation is limited by several methodologic issues. Objective: To investigate the association between recent and chronic prescription-acquired acetaminophen use and asthma. Methods: This retrospective case-control study used a 10% random sample of the IMS LifeLink commercial claims data from 1997 to 2009. Cases had to have at least 1 incident claim of asthma; 3:1 controls matched on age, sex, and region were randomly chosen. Acetaminophen exposure, dose, and duration were measured in the 7- and 30-day (recent) and the 1-year (chronic) look-back periods. Multivariable conditional logistic regression was used to estimate the risk of asthma associated with acetaminophen use adjusted for comorbidities, other drugs increasing asthma risk, and health system factors. Results: There were 28,892 cases and 86,676 controls, with mean age of 42.8 years; 37.7% were mates, and 22.6% of cases and 18.2% of controls had acetaminophen exposure in the pre-index year, with mean cumulative doses of 78.7 g and 59.8 g, respectively. There was no significant association between recent prescription acetaminophen exposure and asthma (7 days: OR 1.02, p = 0.74; 30 days: OR 0.97, p = 0.38). Cumulative prescription acetaminophen dose in the year prior increased asthma risk compared to acetaminophen nonusers (≤1 kg: OR 1.09, p < 0.001 and >1 kg: OR = 1.60, p = 0.02). Duration of prescription acetaminophen use greater than 30 days was associated with elevated asthma risk (OR 1.39, p < 0.001). Conclusions: Chronic prescription-acquired acetaminophen use was associated with an increased risk of asthma, while recent use was not. However, over-the-counter acetaminophen use was not captured in this study and further epidemiologic research with complete acetaminophen exposure ascertainment and research on pathophysiologic mechanisms is needed to confirm these relationships.

Towards an ontological theory of substance intolerance and hypersensitivity

A proper ontological treatment of intolerance--including hypersensitivity--to various substances is critical to patient care and research. However, existing methods and standards for documenting these conditions have flaws that inhibit these goals, especially translational research that bridges the two activities. In response, I outline a realist approach to the ontology of substance intolerance, including hypersensitivity conditions. I defend a view of these conditions as a subtype of disease. Specifically, a substance intolerance is a disease whose pathological process(es) are realized upon exposure to a quantity of substance of a particular type, and this quantity would normally not cause the realization of the pathological process(es). To develop this theory, it was necessary to build pieces of a theory of pathological processes. Overall, however, the framework of the Ontology for General Medical Science (which uses Basic Formal Ontology as its uppermost level) was a more-than-adequate foundation on which to build the theory.