Automated screening of patient electronic medical records is only the first step to identifying a medication problem
Without in-depth analysis of possible adverse drug events (ADEs) after they have been identified from patients' electronic medical records (EMRs), changes to prevent future ADEs is unlikely, a new study concludes. The researchers found that trigger tools, software that identifies possible ADEs (such as giving an infant naloxone to counteract oversedation with opiates, or giving a pediatric patient intravenous glucose while on insulin, indicating insulin-induced hypoglycemia), find more ADEs than voluntary reporting systems. However, in-depth analysis of each possible ADE is necessary to find patterns that can be addressed by changes in clinical procedures.
In the study, conducted over 19 months at Cincinnati Children's Hospital Medical Center, the software identified 50 triggers representing 30 instances of insulin-related hypoglycemia ADEs and 59 triggers representing 34 opiate-related oversedation ADEs. Only 2 of the hypoglycemia ADEs and 3 of the oversedation ADEs had been reported to the hospital's voluntary safety reporting system within 24 hours of the event.
Investigation of characteristics of the insulin-related ADEs, including 53 percent in patients without diabetes who were being treated for hyperglycemia, led the researchers to conclude that the most likely causes were lack of standardization in insulin-dosing decisions, and that patients receiving continuously infused insulin were at highest risk. Similar analysis of the oversedation ADEs indicated that 50 percent of the 34 events occurred on the night shift, and among the 19 oversedated patients, all had their ADEs within 48 hours of surgery.
The findings were based on a database that stored information collected on each trigger event from the medical record and from interviews with frontline staff. The study was funded in part by the Agency for Healthcare Research and Quality (HS16957).
More details are in "Identifying causes of adverse events detected by an automatic trigger tool through in-depth analysis," by Stephen E. Muething, M.D., Patrick H. Conway, M.D., M.Sc., Elizabeth M. Kloppenborg, M.S.N., B.S.N., and others in the October 2010 Quality and Safety in Health Care 19(5), pp. 435-439.
Return to Contents
Proceed to Next Article