- Better Patient Care: Virtually There
- Event Log Management & Compliance Best Practices: For Government & Healthcare Industry Sectors
- The State of EHR Adoption: On The Road to Improving Patient Safety
- Your Cloud in Healthcare - How to Use the Cloud to Achieve Greater Business Agility
- HIE Interoperability case study: Health-e-cITi-NJ
In a novel use of electronic health record data, researchers affiliated with Massachusetts General Hospital have found evidence supporting recent Food and Drug Administration (FDA) warnings of the arrhythmia risks associated with the common antidepressant medication citalopram, which is sold under the brand name Celexa.
The study, published in the journal BMJ, analyzed the use of citalopram and other antidepressant medications in 38,000 patients, compared to biomarker signs of increased risk for ventricular arrhythmia, an abnormal heart rhythm that can be life threatening.
The FDA first warned of arrhythmia risks associated with citalopram, which is also sold in various generic forms, and last year recommended that doses not exceed greater than 40 mg per day “because it could cause potentially dangerous abnormalities in the electrical activity of the heart.” The FDA also discouraged its use for people with certain conditions, such as underlying heart problems or low potassium and magnesium blood levels.
The drug is thought to be able to interfere with the heart’s electrocycle in some people, potentially prolonging the QT wave interval, which can lead to arrhythmia. To measure that association, a research team led by Partners HealthCare research computer scientist Victor Castro examined EHR data from 38,397 patients who between 1990 and 2011 had received an electro-cardiogram in the 90 days after being prescribed one of 11 antidepressant drugs or methadone, a synthetic opioid often prescribed for heroin addiction withdrawal that’s also associated with prolonged QT intervals.
Using Partner’s i2b2 clinical research software (which is also available as an open source license), the team confirmed the association of “a slight but significant QT prolongation” with higher doses of citalopram, in addition to associations with methadone and with amitriptyline, an older antidepressant. At the same time, the results showed that other common antidepressants — like those sold under the brand names Prozac, Paxil and Zoloft — had no effect on QT interval, and that the antidepressant bupropion (sold as Wellbutrin or Zyban) was associated with a shortened QT interval.
Roy Perlis, MD, a study coauthor and director of the Center for Experimental Drugs and Diagnostics at Massachusetts General Hospital, stressed that the results of the study should not encourage patients taking citalopram to change their medications, unless they have other risk factors for arrhythmias. "It was important to confirm the effects of citalopram,” Perlis said, “because the FDA warning really gave us minimal clinical guidance.”
"I worry more about people stopping their antidepressants unnecessarily than about the QT prolongation risks,” Perlis said in a media release. “The real message to patients taking these drugs is to have a conversation with their doctors."
Funded by the National Institutes of Health, the study used the i2b2 software to run through EKG reports while extracting QT interval and other data. “Doing this by hand – flipping through individual patient charts – would have taken a year or more,” Perlis said. “Doing it with electronic health records took about an hour."
The finding of QT shortening with bupropion, Perlis said, “show how this approach can help us find drugs with unexpected benefits and not just unexpected problems. As long as we're willing to accept the limitations – particularly the fact that people aren't randomly assigned to different treatments – this strategy allows us to study many more patients and get answers much faster.”
And in terms of patient privacy, the EHR-based study was “actually much safer than the older methods which required a person to look through a pile of medical records one by one,” he said. “This way we only extract the data we need and never see anything that would allow us to identify an individual patient."