Automated Retract-and-Reorder Measures to Improve Medication Safety


Supporting Health Systems in Advancing Care Delivery


Using Digital Healthcare Tools to Improve Patient Safety

New measures to identify near-miss medication errors are a major advancement in patient safety and can help healthcare systems make ordering even safer.

Near-miss medication errors are difficult to identify and are underreported but can be opportunities to make electronic ordering safer

Medication errors are the most common and preventable cause of patient harm. Yet efforts to prevent these errors have been hampered by lack of standardized measures. Near-miss medication errors, such as when clinicians realize they’ve ordered the wrong dose or frequency for a drug, are caught before the error ever reaches the patient. Although patient harm may have been avoided, understanding the circumstances surrounding near-miss errors presents opportunities to improve safe ordering practices. Research on electronic ordering systems and processes can provide insight into contributors to medication errors, as well as insight into potential solutions.

In previous research, Dr. Jason Adelman of Columbia University developed and validated the automated Wrong-Patient Retract-and-Reorder (RAR) Measure to identify wrong-patient electronic orders. The Wrong-Patient RAR measure identifies orders placed for a patient that are retracted within 10 minutes, and then placed by the same clinician for a different patient within the next 10 minutes. These near-miss errors are self-caught by the clinician before they reach the patient. The Wrong-Patient RAR measure enabled systematic and objective identification of wrong-patient orders in electronic health record (EHR) data, resulting in a critical breakthrough in patient safety research. As Dr. Adelman noted, “Before the development of the Wrong-Patient RAR measure, there were [approximately] six voluntary reported wrong-patient order errors each year at NewYork-Presbyterian versus 10,000 errors identified in a year after the Wrong-Patient RAR measure was implemented. You can't study errors and improve systems when there are only six events.”

Expanding the RAR methodology to other medication order errors further improves understanding and safety

Based on this pioneering work, Dr. Adelman wanted to expand RAR measures to medication and other types of order errors. Using the RAR methodology, Dr. Adelman and his team developed new measures to capture instances where an order was placed, retracted by the ordering clinician, and subsequently reordered by the same clinician for the same patient with a parameter of the order changed. These measures identify electronic order errors including wrong dosage, wrong route, wrong frequency, or wrong medication.

A unique aspect of the research is that the team was able to query for RAR events every 30 minutes and then contact clinicians within 6 hours of each RAR event to understand what happened, including why the initial order was placed, why the order was canceled, who prompted the cancelation, and why the medication was reordered with a parameter changed (e.g., dose, frequency, route). This allowed the researchers to classify whether the event was a true error or not, and to better understand contributing factors for these different errors.

“This research allows us to examine the epidemiology of these errors and identify targets for intervention. Are there differences in frequency by type of errors, for example, wrong-patient versus wrong-dose errors? Are there differences by shift, for example, when you’re tired at night versus during the day? Are there differences between attendings and house staff? You can get very accurate data without the biases of chart review or voluntary reporting.”- Dr. Jason Adelman

RAR measures help identify errors and test interventions to prevent them

The RAR measures were studied at seven hospitals with over 3,000 inpatient beds and six emergency departments with two different EHR systems. The team validated the measures by calculating their positive predictive value (PPV), which tests how well the measures detected true errors. All measures achieved a high PPV, reaching the target of greater than 75 percent. The research showed that automated measurement of electronic order errors can be readily integrated into health system EHRs to study the epidemiology of order errors and to test the effectiveness of proposed EHR improvements on order error outcomes. Dr. Adelman and his team will publicly share the specifications for the RAR measures developed in this study so that the measures can be readily and widely used by other healthcare systems.