Evaluation of drug-drug interaction in the EHR data

Drug–drug interactions (DDIs) caused by co-prescriptions can alter the effectiveness of treatment or occurrence of adverse drug events (ADEs). Also, it in itself could be life-threatening to patients. An electronic alert for DDI is one of the most successful components of electronic medical record (EMR) systems for reducing the DDI-associated ADEs by providing alerts and/or detailed information at the moment of prescribing. The primary objectives of this study is to determine whether the incidence of alerts for DDI and their override rates differed by admitting department (emergency department, ED; general ward, GW; intensive care unit, ICU). We also tried to reveal the difference in drug pairs causing DDI alerts and reasons for alert overrides among the three departments by analyzing a longitudinal electronic health record (EHR) data and its alert and logging system of a teaching hospital. The risk of DDI among admitting departments were compared and analyzed the cause of prescribing which is possible to occur DDI with the purpose of providing baseline knowledge for acceptable clinical decision support system (CDSS). This is a retrospective cohort study using a tertiary hospital in Korea. The participants of this study are all patient who were visited/admitted into ED, GW or ICU, from September 1, 2009 to March 31, 2013, incidentally. The risk for DDI alert and alert overrides were compared among admitting department (ED, GW and ICU). Also, the cause of prescribing possible DDI drugs was analyzed by manual review of all free-text record recorded by physicians. The incidence of DDI alert is 6.4% (0.13 person-year) in total, order by 18.4% in ICU, 6.3% in ED and 6.2% in GW, respectively. After adjusted in age, gender, age-adjusted Charlson’s comorbidity index (CCI), administration of chemotherapeutic drug, mean number of drug per day, and transfer from other department, the risk of DDI alerts in ED (HR, 1.923; 95% CI, 1.823-2.028) and in ICU (HR, 1.845; 95% CI, 1.635-2.083) are higher than in GW (p-value<0.001). The alert overrides in ED is higher than in GW (HR, 13.007; 95% CI, 10.800-15.664), but in ICU is lower than in GW (HR, 0.805; 95% CI, 0.647-1.001) (p-value<0.001) after adjusting the confounders. In ED and GW, co-prescriptions within ‘anti-inflammatory and antirheumatic products’ are most frequent (90.6% in ED, 43.8% in GW), on the other hand, co-prescriptions of ‘cardiac therapy’ drug and drug for obstructive airway disease are the most frequent in ICU, by categorized using three digits of ATC code. As the results of analysis on the cause of alert overrides from free-text record, ‘clinically irrelevant alerts’ is 80.2% of ED, ‘benefit assessed to be greater than the risk’ is 49.0% in GW and 57.5% in ICU. In conclusion, the DDI alerts and the reasons for alert overrides are different by admitting department. There may be needed different strategies for each department to reduce DDI alerts and alert overrides. Improving alerting rule for discharge medication may reduce 2/3 of total inappropriate alerts in ED. Sophisticated DDI rule-updates considering the patient’s clinical information may be required for GW to make the alerting system acceptable to prescribers.

Author dc:creator

• 안, 은경

Contributors dc:contributor

• 박, 래웅
• 대학원 의학과
• 104994

• 약물상호작용
• 임상의사결정지원시스템
• 전자의무기록
• 내원부서
• 약물역학
• 경고 피로
• 경고 무시
• Drug-drug interaction
• Clinical decision support system
• Electronic medical record
• Admitting department
• Pharmaco-epidemiology
• Alert fatigue
• Alert overrides

Language dc:language

ko

Identifier

http://dcoll.ajou.ac.kr:9080/dcollection/jsp/common/DcLoOrgPer.jsp?sItemId=000000019067
000000019067

OAI identifier oai:identifier

oai:repository.ajou.ac.kr:201003/11869