This was the first study to quantitatively test the use of SEIPS (Systems Engineering Initiative for Patient Safety) model to identify factors influencing a medication safety outcome. By using a SEIPS model, our study developed a comprehensive approach to identifying potential factors influencing adverse drug events (ADEs). The SEIPS work system is composed of five components which include person, organization, technologies and tools, tasks, and environment. SEIPS model successfully explained the work system factors influencing ADEs and falls in nursing homes (NHs). The second important contribution of our study is that it used the CMS (Centers for Medicare and Medicaid Services) ADE Trigger Tool not only to detect actual ADEs, but also to identify specific potential ADEs in NHs.
This study had five objectives: 1) calculate actual ADE incidence rate (number of incidents per 100 residents per month) in NHs using the ADE trigger tool, 2) measure potential ADE incidence rate based on abnormal lab data, vital signs and non-harmful falls, 3) identify the classes of medications most likely to cause ADEs, 4) evaluate the relationships between work system characteristics and the incidence of ADEs, and 5) assess the relationships between work system characteristics and resident fall incidents.
This study was an observational quantitative study. It included two quantitative methods: retrospective resident medical chart extraction and survey four types of healthcare practitioners. The staff surveys included four categories of NH practitioners at each facility to ensure comprehensive assessment of the work system: Director of nursing (DON), registered nurse (RN), certified nurse assistant (CNA) and consultant pharmacist. The surveys included questions about the facility conditions, environment, technology, task, and staff/practitioners. Both methods were conducted within the same facilities and during the same period. The study was conducted in 11 NHs in nine cities in Iowa. Data collection was conducted over fall 2016 and spring 2017. Binary logistic regression with Generalized Estimated Equation (GEE) was used to measure the association between the ADE incidence (Yes/No) and characteristics of residents and facilities. The secondary outcome was the incidence of falls.
We reviewed 755 medical charts and conducted 44 staff surveys. The rate of ADEs was 6.13 incidents per 100 residents per month. Approximately (64.1%) of the ADEs were preventable. More than half of the ADEs were fall-related (51.1%) and half of those harmful falls were due to hypotension. We considered all the harmful falls as ADEs in residents with one or more psychotropic, antihypertensive, opioid and/or anti-diabetic medications, which can cause fall. The most common ADEs included medication (opioid)-induced constipation (24.6%), psychotropic induced confusion, dizziness or drowsiness (6.5%), antibiotic-induced Clostridium difficile diarrhea (4.2%), anticoagulant induced bleeding (3.9%) and antidiabetic induced hypoglycemia (3.2%). The most common fall-related ADEs were bruise (9.7%) and abrasion or laceration (9.4%). Psychotropic medications (74.9%), antidepressants (61.3%), antihypertensive agents (58.7%), and opioids (51.9%) were the most common medications associated with ADEs. The rate of potential ADEs was 48.6 per 100 residents per month. The rate of falls was 23.38 per 100 residents per month.
The regression analysis revealed significant associations between the ADEs and opioid analgesics, psychotropic medications, warfarin, skilled care, consultant pharmacist accessibility, nurse-physician collaboration, CNA skills in taking vital signs, number of physician visits to the facility, nurse workload and the use of electronic health records. On the other hand, the regression analysis showed non-significant relationships between ADEs and cardiac arrhythmia (AFib), DON years in the facility and distracting noise during medication administration. The six significant facility characteristics represent five concepts of the SEIPS model: organization, task, environment, person and technology. In the fall regression analysis, twelve of the resident and the facility SEIPS variables had significant relationships with the incidence of resident falls. The significant variables represent four concepts of the SEIPS model: organization, task, environment, and person. Longer DON years in the facility and more nurse time per resident per day were associated with lower number of fall incidents. The CNA skills in taking vital signs have significant negative association with both ADEs and falls. Finally, the variable “CNAs work fast” and the nurse workload also have positive association with the incidence of falls