Automated detection of wrong-drug prescribing errors.

BACKGROUND: To assess the specificity of an algorithm designed to detect look-alike/sound-alike (LASA) medication prescribing errors in electronic health record (EHR) data. SETTING: Urban, academic medical centre, comprising a 495-bed hospital and outpatient clinic running on the Cerner EHR. We extracted 8 years of medication orders and diagnostic claims. We licensed a database of medication indications, refined it and merged it with the medication data. We developed an algorithm that triggered for LASA errors based on name similarity, the frequency with which a patient received a medication and whether the medication was justified by a diagnostic claim. We stratified triggers by similarity. Two clinicians reviewed a sample of charts for the presence of a true error, with disagreements resolved by a third reviewer. We computed specificity, positive predictive value (PPV) and yield. RESULTS: The algorithm analysed 488 481 orders and generated 2404 triggers (0.5% rate). Clinicians reviewed 506 cases and confirmed the presence of 61 errors, for an overall PPV of 12.1% (95% CI 10.7% to 13.5%). It was not possible to measure sensitivity or the false-negative rate. The specificity of the algorithm varied as a function of name similarity and whether the intended and dispensed drugs shared the same route of administration. CONCLUSION: Automated detection of LASA medication errors is feasible and can reveal errors not currently detected by other means. Real-time error detection is not possible with the current system, the main barrier being the real-time availability of accurate diagnostic information. Further development should replicate this analysis in other health systems and on a larger set of medications and should decrease clinician time spent reviewing false-positive triggers by increasing specificity.

Identifying health information technology related safety event reports from patient safety event report databases.

OBJECTIVE: The objective of this paper was to identify health information technology (HIT) related events from patient safety event (PSE) report free-text descriptions. A difference-based scoring approach was used to prioritize and select model features. A feature-constraint model was developed and evaluated to support the analysis of PSE reports. METHODS: 5287 PSE reports manually coded as likely or unlikely related to HIT were used to train unigram, bigram, and combined unigram-bigram logistic regression and support vector machine models using five-fold cross validation. A difference-based scoring approach was used to prioritize and select unigram and bigram features by their relative importance to likely and unlikely HIT reports. A held-out set of 2000 manually coded reports were used for testing. RESULTS: Unigram models tended to perform better than bigram and combined models. A 300-unigram logistic regression had comparable classification performance to a 4030-unigram SVM model but with a faster relative run-time. The 300-unigram logistic regression model evaluated with the testing data had an AUC of 0.931 and a F1-score of 0.765. DISCUSSION: A difference-based scoring, prioritization, and feature selection approach can be used to generate simplified models with high performance. A feature-constraint model may be more easily shared across healthcare organizations seeking to analyze their respective datasets and customized for local variations in PSE reporting practices. CONCLUSION: The feature-constraint model provides a method to identify HIT-related patient safety hazards using a method that is applicable across healthcare systems with variability in their PSE report structures.

Cognitive tests predict real-world errors: the relationship between drug name confusion rates in laboratory-based memory and perception tests and corresponding error rates in large pharmacy chains.

BACKGROUND: Drug name confusion is a common type of medication error and a persistent threat to patient safety. In the USA, roughly one per thousand prescriptions results in the wrong drug being filled, and most of these errors involve drug names that look or sound alike. Prior to approval, drug names undergo a variety of tests to assess their potential for confusability, but none of these preapproval tests has been shown to predict real-world error rates. OBJECTIVES: We conducted a study to assess the association between error rates in laboratory-based tests of drug name memory and perception and real-world drug name confusion error rates. METHODS: Eighty participants, comprising doctors, nurses, pharmacists, technicians and lay people, completed a battery of laboratory tests assessing visual perception, auditory perception and short-term memory of look-alike and sound-alike drug name pairs (eg, hydroxyzine/hydralazine). RESULTS: Laboratory test error rates (and other metrics) significantly predicted real-world error rates obtained from a large, outpatient pharmacy chain, with the best-fitting model accounting for 37% of the variance in real-world error rates. Cross-validation analyses confirmed these results, showing that the laboratory tests also predicted errors from a second pharmacy chain, with 45% of the variance being explained by the laboratory test data. CONCLUSIONS: Across two distinct pharmacy chains, there is a strong and significant association between drug name confusion error rates observed in the real world and those observed in laboratory-based tests of memory and perception. Regulators and drug companies seeking a validated preapproval method for identifying confusing drug names ought to consider using these simple tests. By using a standard battery of memory and perception tests, it should be possible to reduce the number of confusing look-alike and sound-alike drug name pairs that reach the market, which will help protect patients from potentially harmful medication errors.

Characterizing the pain score trajectories of hospitalized adult medical and surgical patients: a retrospective cohort study.

Pain care for hospitalized patients is often suboptimal. Representing pain scores as a graphical trajectory may provide insights into the understanding and treatment of pain. We describe a 1-year, retrospective, observational study to characterize pain trajectories of hospitalized adults during the first 48 hours after admission at an urban academic medical center. Using a subgroup of patients who presented with significant pain (pain score >4; n = 7762 encounters), we characterized pain trajectories and measured area under the curve, slope of the trajectory for the first 2 hours after admission, and pain intensity at plateau. We used mixed-effects regression to assess the association between pain score and sociodemographics (age, race, and gender), pain medication orders (opioids, nonopioids, and no medications), and medical service (obstetrics, psychiatry, surgery, sickle cell, intensive care unit, and medicine). K-means clustering was used to identify patient subgroups with similar trajectories. Trajectories showed differences based on race, gender, service, and initial pain score. Patients presumed to have dissimilar pain experiences (eg, sickle vs obstetrical) had markedly different pain trajectories. Patients with higher initial pain had a more rapid reduction during their first 2 hours of treatment. Pain reduction achieved in the 48 hours after admission was approximately 50% of the initial pain, regardless of the initial pain. Most patients' pain failed to fully resolve, plateauing at a pain score of 4 or greater. Visualizing pain scores as graphical trajectories illustrates the dynamic variability in pain, highlighting pain responses over a period of observation, and may yield new insights for quality improvement and research.

Avoiding Medication Errors: Reducing Harm in Residents Using Oral Anticoagulants.

Medication errors involving oral anticoagulants have led to serious adverse events, including hemorrhage, treatment failures leading to thromboembolic events, and death. This article will highlight medication errors that may arise during the use of oral anticoagulants and provide risk-reduction strategies to address the potential for error and patient harm.

Reducing harm in patients using insulin.

Controlling blood sugars with insulin is an important part of managing hyperglycemia in diabetic patients. However, the literature has shown that the use of insulin has been associated with more medication errors than any other type or class of drug. This article will highlight medication errors that may arise during the use of insulin in the long-term facility and provide risk-reduction recommendations to address the potential for error and patient harm.

Reducing patient harm with the use of opioid oral solutions.

Medication errors involving oral opioids have led to serious adverse events, including failure to control pain, over-sedation, respiratory depression, seizures, and death. This article will highlight medication errors that may arise during the use of opioid oral solutions, particularly concentrated formulations, and provide risk-reduction recommendations to address the potential for error and patient harm.

Reducing patient harm with the use of fentanyl transdermal system.

Fentanyl transdermal system is indicated for the management of persistent, moderate-to-severe chronic pain that requires continuous opioid administration for an extended period of time and cannot be managed by other means such as nonsteroidal analgesics, opioid combination products, or immediate-release opioids. The Institute for Safe Medication Practices and other patient safety advocates have long been concerned that transdermal fentanyl is often prescribed, dispensed, and administered without proper consideration of patient selection criteria, starting-dose recommendations, contraindications, dose adjustment recommendations, and safe administration procedures. This article will highlight errors that may arise during the use of fentanyl patches and provide risk-reduction recommendations for consultant pharmacists to address the potential for error and patient harm.