Sarah Collins, ’09, PhD, RN, is an assistant professor of biomedical informatics and nursing at Columbia University. She has a joint appointment in the Department of Biomedical Informatics (DBMI) and the School of Nursing and recently returned to Columbia after earning a PhD in nursing informatics at the School of Nursing in 2009 and completing a National Library of Medicine Fellowship at DBMI in 2011. In the interim, she was a senior nurse and clinical informatician at Brigham and Women’s Hospital and Partners Healthcare System and an instructor in medicine at Harvard Medical School.
Dr. Collins’ research focuses on identifying and intervening on system-level weaknesses to improve patient outcomes. Her work has been recognized and awarded by the American Medical Informatics Association (AMIA) and the International Congress on Nursing Informatics, and she was selected as one of MedTech Boston’s 40 Under 40 Healthcare Innovators in 2017.
What inspired you to pursue nursing as a career?
I always knew I wanted to work in health care. When it came time to look at undergraduate programs, I applied to both nursing and non-nursing schools. I chose to attend the University of Pennsylvania and majored in nursing with a minor in health care management because I was impressed by the program and the opportunities for collaboration.
As an undergraduate, I developed a keen interest in cardiac and critical care nursing. I discovered that the science and practice of nursing in this area fit within what I had envisioned but couldn’t previously define in terms of my interests in healthcare.
How did your early clinical experiences as a critical care nurse lead you to pursue a PhD in nursing informatics?
Early in my career, I worked in the cardiac intensive care unit at Massachusetts General Hospital. During these years, I began to understand how information technology (IT) systems and software solutions could make work either easier or more challenging for health care professionals.
After, I went on to practice at another hospital and observed that my ability to deliver care was different. I thought, ‘The types of patients are the same. The best practices and standards of care for those patients are the same. What’s different?’
The answer was the physical environment, the culture, and the IT systems. I realized that the importance of how systems are implemented, including how we capture data from nurses and how we expect them to document patient care, is essential to nurses’ ability to deliver high quality, safe care.
This led me to question, ‘Why isn’t our care always the same, and can we do better?’ These were the questions that led me to pursue research and enter Columbia Nursing’s PhD program.
Tell us what your work is focused on now.
I am co-leading a study at Columbia called CONCERN with colleague Kenrick Cato ’08 ‘14, PhD, assistant professor of nursing. The study, which is funded by a grant from the National Institute of Nursing Research, includes sites at Brigham and Women’s Hospital and Harvard Medical School. It is focused on predictive analytics using various data science methods, primarily of nursing documentation, in the inpatient hospital setting to identify nurses’ concern about a patient as an early warning signal for deterioration.
We know from previous work that there is a signal within nursing documentation, particularly around the patterns and frequency of their documentation, that is a proxy for a nurse’s concern about a patient’s clinical state. We have found this is associated with patient outcomes of mortality and cardiac arrest in the hospital. These signals can be subtle. When a nurse is worried about a patient, they assess them more often. Perhaps a patient’s vital signs are fairly normal, so according to the numbers there doesn’t appear to be a problem. But the nurse, who knows the patient well, sees that the patient doesn’t look as good or their breathing is just a little labored.
We think detecting this surveillance behavior can be useful for increasing awareness of the patient state—otherwise known as situational awareness—among the larger health care team. Additionally, communication in a hospital environment is known to be a major factor in patient safety errors. There are challenges, however, in making all care team members, who may be in different locations, aware of changes in a patient’s clinical status and intervening quickly to mitigate those changes.
Computer systems can be leveraged to facilitate these processes. One of the aims of the study is to work closely with care providers to build a clinical decision support system that can notify team members when a patient has expressed these signals, allowing them to better anticipate patient deterioration and facilitate early intervention. We are using the FHIR (Fast Healthcare Interoperability Resources) standard to implement this clinical decision support in two different electronic health record systems at New York Presbyterian and Partners Healthcare to increase generalizability and scalability.
What brought you back to Columbia after earning your PhD here?
I always loved Columbia Nursing, and the Department of Biomedical Informatics is wonderful. I was very happy and excited to return and to make this my home.