Assessing Prescriber Behavior with a Clinical Decision Support Tool to Prevent Drug-Induced Long QT Syndrome

Trinkley, Katy E.; Pell, Jonathan; Martinez, Dario D.; Maude, Nicola R.; Hale, Gary; Rosenberg, Michael

doi:10.1055/s-0041-1724043

Cited by 8 publications

(14 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This finding points to the multifactorial nature of diLQTS, highlighting the need to consider other relevant contextual factors in assessing risk. However, it may also suggest that in the inpatient setting, there might be more benefit than risk with using these medications, which is also consistent with prior studies [9][10][11][12], including one where a clinical decision support tool to prevent diLQTS had a paradoxical decrease in mortality for patients in whom the treating provider ignored the alert and prescribed the known QT-prolonging medication despite risk [4]. Particularly in subjects who were not critically ill (not in cluster 0) and without a history of cardiovascular disease (not in cluster 1), there appeared to be more benefit to using ondansetron, balanced against more risk with using propofol.…”

Section: Principal Findingssupporting

confidence: 76%

“…Drug-induced long-QT syndrome (diLQTS) [1,2] is a major concern for inpatients worldwide and has been identified as a key target for clinical decision support tools [3][4][5][6][7]. Importantly, although certain medications have been implicated as having significant clinical risk [8,9], for others, despite a known risk of diLQTS, clinical validation has been lacking [10][11][12].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Interpretable Machine Learning Prediction of Drug-Induced QT Prolongation: Electronic Health Record Analysis

Simon¹,

Trinkley²,

Malone³

et al. 2022

J Med Internet Res

View full text Add to dashboard Cite

Background Drug-induced long-QT syndrome (diLQTS) is a major concern among patients who are hospitalized, for whom prediction models capable of identifying individualized risk could be useful to guide monitoring. We have previously demonstrated the feasibility of machine learning to predict the risk of diLQTS, in which deep learning models provided superior accuracy for risk prediction, although these models were limited by a lack of interpretability. Objective In this investigation, we sought to examine the potential trade-off between interpretability and predictive accuracy with the use of more complex models to identify patients at risk for diLQTS. We planned to compare a deep learning algorithm to predict diLQTS with a more interpretable algorithm based on cluster analysis that would allow medication- and subpopulation-specific evaluation of risk. Methods We examined the risk of diLQTS among 35,639 inpatients treated between 2003 and 2018 with at least 1 of 39 medications associated with risk of diLQTS and who had an electrocardiogram in the system performed within 24 hours of medication administration. Predictors included over 22,000 diagnoses and medications at the time of medication administration, with cases of diLQTS defined as a corrected QT interval over 500 milliseconds after treatment with a culprit medication. The interpretable model was developed using cluster analysis (K=4 clusters), and risk was assessed for specific medications and classes of medications. The deep learning model was created using all predictors within a 6-layer neural network, based on previously identified hyperparameters. Results Among the medications, we found that class III antiarrhythmic medications were associated with increased risk across all clusters, and that in patients who are noncritically ill without cardiovascular disease, propofol was associated with increased risk, whereas ondansetron was associated with decreased risk. Compared with deep learning, the interpretable approach was less accurate (area under the receiver operating characteristic curve: 0.65 vs 0.78), with comparable calibration. Conclusions In summary, we found that an interpretable modeling approach was less accurate, but more clinically applicable, than deep learning for the prediction of diLQTS. Future investigations should consider this trade-off in the development of methods for clinical prediction.

show abstract

Section: Principal Findingssupporting

confidence: 76%

Section: Introductionmentioning

confidence: 99%

Interpretable Machine Learning Prediction of Drug-Induced QT Prolongation: Electronic Health Record Analysis

Simon¹,

Trinkley²,

Malone³

et al. 2022

J Med Internet Res

View full text Add to dashboard Cite

show abstract

“…Other studies examining the response to QT‐related CDS have shown a 21% reduction of ordering noncardiac QTc interval–prolonging medications in a cardiac care unit and a 13.9% to 37.8% reduction in administering QTc interval–prolonging medications to patients with a history of significant QTc prolongation (QTc ≥500 ms). 7 , 8 , 9 In comparison, this study yielded a cancellation rate of 10.2% of incoming known TdP risk medications, and existing medications were canceled in 7.6% of the advisories. There may have been some overlap with situations where both incoming and existing medications were canceled in a single advisory, and the total number of existing known risk medications is not known, but the overall cancellation rate in this study appears similar to those in other QT‐related CDS studies despite differences in the populations to whom the alerts applied.…”

Section: Discussionmentioning

confidence: 84%

“… 7 This alert led to a reduction in the prescribing of noncardiac medications with a known risk of TdP in a cardiac care unit and a decrease in the occurrence of excessive QTc prolongation. Two other examples of QTc‐related CDS tools are alerts that appear when clinicians attempted to order a medication with a risk of QTc prolongation or TdP for patients who had a QTc ≥500 ms. 8 , 9 The Mayo Clinic tool triggers on orders for CredibleMeds known or possible risk drugs and effectively reduced the number of high‐risk medications administered to these patients by 13.9%. 3 , 8 The CDS tool implemented by Trinkley et al was triggered by 10 different QTc‐prolonging medications and resulted in the discontinuation of 37.8% of triggered medication orders.…”

mentioning

confidence: 99%

Clinician Responses to a Clinical Decision Support Advisory for High Risk of Torsades de Pointes

Gallo

Heise

Woosley

et al. 2022

JAHA

View full text Add to dashboard Cite

Background Torsade de pointes (TdP) is a potentially fatal cardiac arrhythmia that is often drug induced. Clinical decision support (CDS) may help minimize TdP risk by guiding decision making in patients at risk. CDS has been shown to decrease prescribing of high‐risk medications in patients at risk of TdP, but alerts are often ignored. Other risk‐management options can potentially be incorporated in TdP risk CDS. Our goal was to evaluate actions clinicians take in response to a CDS advisory that uses a modified Tisdale QT risk score and presents management options that are easily selected (eg, single click). Methods and Results We implemented an inpatient TdP risk advisory systemwide across a large health care system comprising 30 hospitals. This CDS was programmed to appear when prescribers attempted ordering medications with a known risk of TdP in a patient with a QT risk score ≥12. The CDS displayed patient‐specific information and offered relevant management options including canceling offending medications and ordering electrolyte replacement protocols or ECGs. We retrospectively studied the actions clinicians took within the advisory and separated by drug class. During an 8‐month period, 7794 TdP risk advisories were issued. Antibiotics were the most frequent trigger of the advisory (n=2578, 33.1%). At least 1 action was taken within the advisory window for 2700 (34.6%) of the advisories. The most frequent action taken was ordering an ECG (n=1584, 20.3%). Incoming medication orders were canceled in 793 (10.2%) of the advisories. The frequency of each action taken varied by drug class ( P <0.05 for all actions). Conclusions A modified Tisdale QT risk score–based CDS that offered relevant single‐click management options yielded a high action/response rate. Actions taken by clinicians varied depending on the class of the medication that evoked the TdP risk advisory, but the most frequent was ordering an ECG.

show abstract

“…1). 15 While medication-related CDS is designed with good intentions, provider response rates are inconsistent with variation among providers [16][17][18][19] and concerns exist that too many alerts within the electronic health record (EHR) may contribute to alert fatigue. [20][21][22] While there are many studies focused on CDS alerts, data are scarce on the efficacy of alerts that are only active for a short time to address a temporary situation, as well as the susceptibility of these short-term alerts to alert fatigue.…”

mentioning

confidence: 99%

Low Efficacy of Medication Shortage Clinical Decision Support Alerts

et al. 2021

View full text Add to dashboard Cite

Objective We examined clinical decision support (CDS) alerts designed specifically for medication shortages to characterize and assess provider behavior in response to these short-term clinical situations. Materials and Methods We conducted a retrospective analysis of the usage of medication shortage alerts (MSAs) that included at least one alternative medication suggestion and were active for 60 or more days during the 2-year study period, January 1, 2018 to December 31, 2019, in a large health care system. We characterized ordering provider behavior in response to inpatient MSAs. We then developed a linear regression model to predict provider response to alerts using the characteristics of the ordering provider and alert frequency groupings. Results During the study period, there were 67 MSAs in use that focused on 42 distinct medications in shortage. The MSAs suggested an average of 3.9 alternative medications. Adjusting for the different alerts, fellows (p = 0.004), residents (p = 0.03), and physician assistants (p = 0.02) were less likely to accept alerts on average compared with attending physicians. Further, female ordering clinicians (p < 0.001) were more likely to accept alerts on average compared with male ordering clinicians. Conclusion Our findings demonstrate that providers tended to reject MSAs, even those who were sometimes flexible about their responses. The low overall acceptance rate supports the theory that alerts appearing at the time of order entry may have limited value, as they may be presented too late in the decision-making process. Though MSAs are designed to be attention-grabbing and higher impact than traditional CDS, our findings suggest that providers rarely change their clinical decisions when presented with these alerts.

show abstract

Assessing Prescriber Behavior with a Clinical Decision Support Tool to Prevent Drug-Induced Long QT Syndrome

Cited by 8 publications

References 31 publications

Interpretable Machine Learning Prediction of Drug-Induced QT Prolongation: Electronic Health Record Analysis

Interpretable Machine Learning Prediction of Drug-Induced QT Prolongation: Electronic Health Record Analysis

Clinician Responses to a Clinical Decision Support Advisory for High Risk of Torsades de Pointes

Low Efficacy of Medication Shortage Clinical Decision Support Alerts

Contact Info

Product

Resources

About