Skip to main content

Can This Model Reduce the Frequency of Medication Errors?

February 15, 2017

A model developed to predict the occurrence of clinically significant medication errors may help pharmacists and other health care professionals focus interventions on high-risk patients, according to a study published in PloS One.

Antipsychotics Overprescribed for Parkinson’s Patients
Optimizing Anesthesia and Improving Patient Care

Drug-related problems represent a significant proportional of emergency department visits and health care expenditures. These problems are most commonly defined as events of circumstance involving drug therapy that actually or potentially interfere with desired outcomes. As such, reducing the occurrence of these events is an important facet of improving health care.

In a study led by, Tri-Long Nguyen, McMaster University (Hamilton, Ontario, Canada), researchers theorized that the successful intervention of high-risk patients may be one method by which institutions could reduce the frequency of medication errors. However, conventional criteria often fail to identify these patients. Therefore, the researchers developed a multivariate model-based strategy to detect these high-risk patients.

For the study, researchers conducted a prospective cohort study on adult patients at a University Hospital Centre. After an analysis of all medication errors, they simulated 5000 randomized controlled trials and compared two clinical decision pathways for intervention: one supported by their new model, and one based on the criterion of age, which is commonly used at most institutions.

Overall, among 1408 patients, 365 experienced at least one clinically significant medication error. Eleven variables were identified using multivariable logistic regression and used to build a predictive model which demonstrated fair performance (c-statistic: 0.72). Major predictors were age and number of prescribed drugs.

When the model made by the researchers was compared the the standard of care model, researchers found that their model improved the interception of potential adverse drug events by 17.5%.

Thus, they concluded that their model has the potential to help providers better avoid medication errors in high-risk patients; although more research must be done on an independent set of patients and the model should be evaluated through a real clinical impact study.—Sean McGuire

Back to Top