On the QT: EMR Drug Alerts That Physicians Value

Patient safety is a major concern for every health professional. However, for cardiologists, the interaction of certain drugs and heart conditions make vigilance against contraindications and complications an added battle. EMRs can help, but only if they work reliably and clinicians observe alerts.

A heart-stopping reaction

Among the many conditions where physicians stand to ensure patient safety through electronic records-guided methods, QTc interval prolongation and subsequent ventricular tachycardia could make an enormous impact. QTc interval prolongation is a serious problem. As the heart’s beat lengthens, it puts a patient at risk for torsades de pointes, a life-threatening polymorphic ventricular tachycardia, ultimately leading to cardiac arrest (Circ Cardiovasc Qual Outcomes 2014; 7[3]:381-390). When patients present at admission with already lengthened QTc intervals, adding certain drugs to the mix puts them at risk for potentially fatal complications.

In an effort to improve patient safety, research teams have been working independently to come up with ways to alert physicians, pharmacists and other care staff to potential problems before they start. But, part of the problem is finding an effective alert that won’t be ignored.

James E. Tisdale, PharmD, BCPS, is a professor with the College of Pharmacy at Purdue University in Indianapolis. He leads one of the teams that developed an alert using an EMR to give staff notice when a patient who presents with QTc interval prolongation is prescribed a medication that could exacerbate the problem. “This all came about because we had some colleagues in the pharmacy department at Methodist Hospital [in Indianapolis] whose clinical opinion was that they were getting a lot of patients with QT prolongation.”

Tisdale and colleagues found QTc interval prolongation occurs in about a quarter of the patients admitted to the cardiac intensive care units [ICUs] in the Indiana University Health System. “A substantial proportion of those patients, despite having QT prolongation at admission, went on to be prescribed QT interval prolonging drugs in the ICU. So, they were at risk and they went on to receive these drugs anyway, which would increase their risk further.” More than 18 percent of patients overall had a QT interval length of over 500 ms (Drug Saf 2012; 35[6]:459-470).

In order to reduce the risks to patients, Tisdale and colleagues looked to the EMR systems for help. “We thought that one way of addressing the problem was to develop a computer decision support system alert, but we needed it to make it specific enough so that it would only fire when there’s a real problem,” Tisdale says. Otherwise, he notes, healthcare professionals are apt to ignore it.

This meant building the alert from the ground up. Tisdale and colleagues analyzed QTc interval prolonging risk factors and developed a risk score scale that could be coded into their alert system.

When a patient, based on the data coded into the system, was at moderate or high risk for QT prolongation and a medication was prescribed that could potentially lengthen QTc intervals, the alerts triggered. Combined, he noted, patients with a moderate to high risk had a “greater than 50 percent chance of developing QT prolongation in the hospital.”

“When a patient comes into the unit, and actually now into the hospital, the computer calculates their risk score for developing QT prolongation. It will determine whether they’re low, moderate or high risk. The computer won’t do anything with that unless the patient is also prescribed a drug that is known to prolong the QT interval,” Tisdale notes. “When that happens, a computer alert will fire if the patient is at moderate or high risk, but it will not fire if the patient is at low risk. It discriminates. It only fires if there’s a certain level of risk.”

The ability for the system to discriminate between high and low risk is a major step toward improved patient safety. Without it, this system would run into problems experienced by users of most commercial drug interaction alert systems: alert fatigue.

The boy who cried wolf

Alerts provided by many commercial systems are intended to help reduce patient risk, but tend to get ignored. “The problem with those alerts is that they’re nonspecific, usually for legal reasons they’ll fire for any potential of drug interaction, no matter how low the likelihood is or how low the risk of harm is,” Tisdale says. “It’s like the boy crying wolf. People ignore them. They override them. This blindness to alerts is alert fatigue, and it makes what should be a tool for patient safety less of a helpmate and more of a hindrance.”

This is an issue of greatest concern to colleagues Hardeep Singh, MD, MPH, and Dean F. Sittig, PhD. Singh is chief of the Health Policy, Quality and Informatics Program at Michael E. DeBakey Veterans Affairs Medical Center in Houston, while Sittig is a professor of medicine with the school of biomedical informatics at the University of Texas – Memorial Hermann Center for Healthcare Quality and Safety in Houston. They have coauthored several papers on the subject of patient safety and EMRs.

They agree that alerts and EMR systems can improve patient safety, but only when done in a way that doesn’t overload clinicians with irrelevant information. “We’ve shown that clinicians who are overwhelmed with information are missing more abnormal test results because of the information overload,” says Singh.

He agrees a directed approach would help keep extraneous and unnecessary alerts from distracting clinicians and avoid the problems encountered with physicians dismissing alerts out of hand. Sittig adds, “You can only say to a doctor so many times that you did the wrong thing when the computer’s wrong before they tune you out.”

Sittig suggests considering an alert as decision support of last resort. “You want to make the right thing to do the easiest thing to do,” says Sittig. “An alert is like slapping someone on the wrist saying you did it wrong.”

Since alerts are only part of the problem, Sittig and Singh recommend organizations using EMRs review the SAFER guides to help assess the institution’s electronic system. The SAFER guides are intended to help recommend best practices and optimize safe and effective use of EMR systems. Singh, Sittig and their colleagues developed these guides using what they call a sociotechnical approach to better understand where failure is occurring and improve the interaction between human and electronic systems.

“We found that sometimes the computer’s not very good at displaying information, sometimes there’s too much information, sometimes people don’t know how to use it,” says Sittig.

“We’ve got the knowledge, now we need to put it into the systems,” Singh says.

Clear risk, by the numbers

In the case of the algorithm and alerts developed by Tisdale and his team, users only see the alert when particular criteria are met. Users see not only the patient’s risk score but also a list of the patient’s risks. This provides the physician and the pharmacist with a better understanding of what’s going on, the opportunity to modify risk factors and allows them to consider whether another treatment might be necessary. Or, to pay extra attention to monitoring the patient for QT intervals.

Sittig says, “That system, we feel, is better than commercially available drug interaction alerts, because it’s specific, rather than firing for everybody who’s on a drug that might cause a problem.”

Electronic Prescribing Makes Dollars and Sense

The benefits of an e-prescription system may be legion—in accurately managing drugs and their interactions, in saving patients from harm and saving medical systems money.

“One of the areas that we’ve been doing a lot of research in is looking at medication safety, appropriate use of medication and medication errors and the role that electronic medication management systems might have in addressing those issues,” says Johanna Westbrook, PhD. Westbrook is the director for the Centre for Health Systems and Safety Research at Macquarie University in Sydney, New South Wales.

The research, Westbrook says, is part of a larger study involving the impact of electronic medication management systems on major teaching hospitals. Australia has only recently begun implementing EMR systems. As a publicly funded healthcare system, not only could EMRs help improve patient safety and care, it could affect costs.

Westbrook and colleagues noted that when adverse drug events were prevented, significant savings resulted. Patients in the cardiology ward where e-prescribing was implemented experienced a 71 percent decrease in the number of adverse events. And while the costs were around $48,818 (A$61,741) per year for implementing, the savings per admission were between $56 and $59 (A$63–$66). “Which in a year for this ward we would be looking at $100K a year,” she notes. Ultimately, estimated savings for the whole hospital, once implemented, could be significant, around $2 million (approximately A$2.5 million) (J Am Med Inform Assoc online Feb. 10, 2015).

“Australian hospitals are really still in the early stages of introducing electronic prescription systems,” Westbrook notes. “They’re still learning a lot about decision support and how effective that is. I think that’s one area where we can get a lot more benefits: by designing effective decision support.”

Annual Operating Costs Associated with E-system Incurred at the Ward Level

Therapeutic Guidelines30 beds85625,680
Subscription to online reference texts for the integrated clinical decision-support system
Jan-June 201213,500322a
Australian Medicines Handbook14,300396a
Database maintenance and training
(salaries of pharmacists, clinical information
manager and eMMS trainer)
Full time equivalent
One IT Specialist0.25 FTE180,0003,409b
One Clinical Information System Manager
(Health Service Manager, Level 2)
0.5 FTE110,8764,200b
One Senior Pharmacist Grade 3, second year1 FTE110,9246855b
One Clinical Information Trainer
(Health Service Manager, Level 1)
1 FTE90,4828403b
Total operating ward-specific costs50,277  
(a) Allocated in proportion to the number of beds in the ward relative to the total number of beds at the hospital (30/326);
(b) allocated in proportion to “weights” reflecting resources required for the complexity of management of MedChart in different ward environments. Source: J Am Med Inform Assoc online Feb. 10, 2015