Extracting drug indication information from structured product labels using natural language processing.

OBJECTIVE: To extract drug indications from structured drug labels and represent the information using codes from standard medical terminologies. MATERIALS AND METHODS: We used MetaMap and other publicly available resources to extract information from the indications section of drug labels. Drugs and indications were encoded by RxNorm and UMLS identifiers respectively. A sample was manually reviewed. We also compared the results with two independent information sources: National Drug File-Reference Terminology and the Semantic Medline project. RESULTS: A total of 6797 drug labels were processed, resulting in 19 473 unique drug-indication pairs. Manual review of 298 most frequently prescribed drugs by seven physicians showed a recall of 0.95 and precision of 0.77. Inter-rater agreement (Fleiss κ) was 0.713. The precision of the subset of results corroborated by Semantic Medline extractions increased to 0.93. DISCUSSION: Correlation of a patient's medical problems and drugs in an electronic health record has been used to improve data quality and reduce medication errors. Authoritative drug indication information is available from drug labels, but not in a format readily usable by computer applications. Our study shows that it is feasible to use publicly available natural language processing resources to extract drug indications from drug labels. The same method can be applied to other sections of the drug label-for example, adverse effects, contraindications. CONCLUSIONS: It is feasible to use publicly available natural language processing tools to extract indication information from freely available drug labels. Named entity recognition sources (eg, MetaMap) provide reasonable recall. Combination with other data sources provides higher precision.

Computerized physician order entry of medications and clinical decision support can improve problem list documentation compliance.

OBJECTIVE: The problem list is a key and required element of the electronic medical record (EMR). Problem lists may contribute substantially to patient safety and quality of care. Physician documentation of the problem list is often lower than desired. Methods are needed to improve accuracy and completeness of the problem list. DESIGN: An automated clinical decision support (CDS) intervention was designed utilizing a commercially available EMR with computerized physician order entry (CPOE) and CDS. The system was based on alerts delivered during inpatient medication CPOE that prompted clinicians to add a diagnosis to the problem list. Each alert was studied for a 2-month period after implementation. MEASUREMENTS: Measures included alert validity, alert yield, and accuracy of problem list additions. RESULTS: At a 450 bed teaching hospital, the number of medication orders which triggered alerts during all 2-month study periods was 1011. For all the alerts, the likelihood of a valid alert (an alert that occurred in patients with one of the predefined diagnoses) was 96+/-1%. The alert yield, defined as occuring when an alert led to addition of a problem to the problem list, was 76+/-2%. Accurate problem list additions, defined as additions of problems when the problem was determined to be present by expert review, was 95+/-1%. CONCLUSION: The CDS problem list mechanism was integrated into the process of medication order placement and promoted relatively accurate addition of problems to the EMR problem list.

Automating the maintenance of problem list documentation using a clinical decision support system.

We designed and tested a clinical decision support system (CDSS) prototype to investigate whether a CDSS that assists matching ordered drugs to problems on the problem list can enhance the maintenance of medications and problem lists in the electronic medical record. We evaluated the capability of this CDSS in medication-problem matching using clinical expert chart audits. The analysis revealed that this CDSS could determine the completeness of medication and problem lists if a mismatch occurs.

Assessing physician comprehension of and attitudes toward problem list documentation.

We conducted an online survey to assess the knowledge, attitudes, and practice patterns of physicians related to issues in problem list documentation. Respondents felt that a decision support tool to improve problem list documentation would benefit patient safety more than physician productivity. The majority of respondents are reluctant to maintain medication and problem lists and the quality of documentation remains inadequate.

Overcoming limitations of data entry for the semi-automated detection of drug orphans in the EMR.

Sophisticated decision support systems (DSS) can reduce preventable medical errors. A standalone DSS prototype was built to identify drug-disease mismatches in the electronic medical record (EMR). When drugs fail to match a known problem on the problem list (drug orphans), either the problem list is deficient or the drug was ordered in error. We tested the performance of an integrated DSS prototype by improving the data exchange with the standalone DSS prototype. By implementing a screen capture tool, we were able to accelerate data entry into the DSS prototype through the semi-automated operation. Preliminary results revealed a marked increase in the rate of data entry during testing the DSS prototype.The accelerated data entry streamlines workflow and promotes physician's acceptance of the DSS.

Making the neurology clerkship more effective: can e-Textbook facilitate learning?

OBJECTIVE: In 1998, the 4-week neurology elective clerkship was converted into a 2-week required neurology rotation at the University of Illinois at Chicago. We hypothesized that the interactive e-Textbook, a computer-assisted learning tool, could successfully replace a paper-based syllabus and a traditional neurology textbook during a 2-week rotation, while incorporating department teaching conferences to replace the medical student lecture series. METHODS: We created an e-Textbook and made it available simultaneously in a CD-ROM format and on a password-protected website. The online quiz and course assessment were administered by the Blackboard Web Server. RESULTS: After implementation of the e-Textbook over 6 years, the feedback shows high student satisfaction, and student evaluations of the neurology clerkship have risen. Creation of an e-Textbook for the neurology clerkship made our faculty more productive while increasing student satisfaction and facilitating learning efficacy. DISCUSSION: The results show that the e-Textbook is an appropriate alternative to facilitate learning of basic and clinical neurology during a 2-week rotation. The students demonstrated successful learning in a computerized environment.

Mental Status Expert: an expert system for scoring and reporting the mental status examination.

For neurology clerks and residents, the mental status examination is often the most difficult part of the neurological examination. We have developed the Mental Status Expert (MSE), a structured mental status examination designed for use by neurology student clerks and neurology residents. The MSE assesses seven areas of cognitive function: orientation and concentration, language, right parietal functions, left parietal functions, occipital lobe functions, frontal lobe functions, and memory. Norms for the test have been created based on administration to 281 subjects referred to the Neurology Clinic. The MSE uses production rules and other expert system features to assist in the scoring and reporting of test results.

Evaluating a digital Resident Diagnosis Log: reasons for limited acceptance of a PDA solution.

Personal digital assistants (PDAs) offer opportunities to speed the process of data collection and analysis. We developed a PDA-based program, the Resident Diagnosis Assistant, which allows neurology residents to collect key diagnostic information about patients seen and treated and the upload this data into a departmental computer. Physicians often use the expenditure of their time to decide whether to accept a new device.