Meet Colleen Karvetski, PhD – Sr. Data Scientist at Cigna

By admin  |  March 11, 2020  |  Other

Dr. Colleen

In support of the care team, data and education provide  the foundation for patient safety. Data analysis identifies the critical areas that require focus, proves or disproves hypothesis, creates a baseline for improvement, and highlights the outcomes. In observance of Patient Safety Awareness Week, we spoke to industry leaders to get their thoughts on the challenges to patient safety and how care teams can overcome them.

A veteran in healthcare analytics and senior data scientist at Cigna, Colleen Karvetski, PhD shares her experience and the role of the data scientist in patient safety initiatives.

I know that your educational background includes a doctoral degree in systems engineering. How did you happen to bring your education and background to become a data scientist in healthcare?

After completing a master’s degree in mathematical sciences, I decided to pursue a PhD with a more applied focus, and chose systems engineering, which is essentially applied math focused on modeling everything from human metabolism to human decision making. I met a group of mathematics professors that were searching for graduate student research assistants to support NIH and NIDDK-grant funded research dedicated to optimizing insulin dosing for patients with type 1 diabetes. I feel very fortunate to have found the team – I loved mathematics and now I had an opportunity to apply it in a way that could directly improve quality of life for people with type 1 diabetes. After completing my PhD and a post doc, I was drawn to the health care system setting. I spent 5 years as a data scientist supporting the critical care service line in a large health care system; I am now working in the health insurance industry to support the optimal identification, engagement, and support for patients in clinical programs designed to advocate for them on their health care journey.

In the world of patient safety and quality improvement in healthcare systems, clinicians are familiar with hearing about a “team approach.” But the term “team” is most often limited to other clinicians- a table filled with physicians, nurses, advanced practice providers and pharmacists from various hospital departments. I know that you currently work on the payer side, but from your experience in a hospital setting, can you give me an example of how data scientists contribute to patient safety initiatives?

In the context of patient safety (PS) initiatives, the role of a data scientist is really to support the health care team in understanding and measuring the initiative’s value. To be effective, it is critical that the data scientist be considered as a member of the interdisciplinary team, ideally from the beginning of the initiative design process. The data scientist can support the team in various steps of a PS initiative:

First, understanding the problem and the key drivers – before deploying a new initiative (or measuring an existing one to assess its effectiveness), it is a helpful exercise to understand the problem and the key drivers associated with the patient safety measure that the initiative seeks to address. Patient safety measures themselves are complicated – understanding how a given measure is defined (for example, which patients are considered for the measure, how the measure is defined based on documentation, and the risk adjustment methodology applied in the calculation of the measure). Underperformance on a patient safety measure may not be a result of suboptimal care, but rather the way in which the care or other related measure criteria is documented. A data scientist can help to assess this. If there are opportunities to improve patient safety through an initiative, a data scientist can help in identifying the greatest opportunity for improvement.

(A little more on the concept of risk adjustment — in the case of patient safety indicators (PSIs) and other publicly reported measures captured by CMS, risk adjustment is almost always conducted. This means that CMS considers baseline factors that could influence the PSI that cannot be controlled by the care that is administered. CMS adjusts for these factors so that hospitals or health systems are not penalized for having patients that may be inherently higher risk. A data scientist can help the team to understand what these factors are and whether they are being documented/captured appropriately.)

Once a measure is understood and the PS initiative to impact the measure is defined, a data scientist can assist in the design and deployment of the initiative. The design of the initiative should facilitate conclusive, action-oriented results. Considerations in the design phase include:

Can patients be randomized to receive the new intervention vs not? This is sometimes ethically not feasible, but the gold standard for any intervention design is randomization (variations on this theme could be randomization at the hospital or office level). In the absence of randomization, a pre/post design can also be used; that is, comparing outcomes before the intervention vs after.

Does the team have the mechanisms in place necessary to collect the data to measure the initiative’s effectiveness? For example, if a new handwashing protocol is identified as an opportunity to improving post-operative sepsis rates, can compliance with the new protocol be measured in an accurate way?

How long will the new process run before it will be considered a success or not? And ultimately, how will success be defined? This should be decided a priori. This way, when the initiative concludes and the results are analyzed, the findings will be conclusive and action-oriented.

A data scientist should most importantly assist in providing solutions and recommendations based on the findings from the analysis performed.

What advice would you give a data scientist new to assuming a role in a healthcare system?

The best advice I can give is to build relationships with the interdisciplinary team that you will be supporting. Understand the clinical workflow and immerse yourself in the clinical world as much as you are able. Go on rounds with physicians and tour the hospital units or physician offices. If you are supporting a sepsis initiative, visit a sepsis patient and look over the shoulder of a clinician as they care for the patient and document information in the patient’s chart. Building these relationships with clinical leaders will benefit you immensely, and understanding the patient care process from the front lines will inspire and bring immeasurable value to the work that you do. Because you’ve taken the time to learn their world, the clinical team will respect your feedback and learn to trust in the information that you provide back to them. I cannot express enough how important it is to establish mutual trust and respect with your stakeholders. Most of the requests you will receive in your role will come from health care leaders. But my advice would be to be proactive in proposing data driven solutions. As you become more familiar with clinical process and workflow, don’t be afraid to make suggestions on what kind of analysis or modeling/information would be meaningful in supporting the frontline staff and benefitting patient care.

What advice would you give to a physician or nurse leader working with you on a project team who has not previously collaborated with a data scientist?

The more that a physician or nurse leader can invest in making the data scientist a part of the interdisciplinary team, the more value that the data scientist will be able to offer to the team in return. The data scientist relies on the clinical team to teach them about the disease process, the clinical workflow, and the goals of the initiative (and of course the political climate/culture that may have an impact on the success of a new intervention or process!). By being integrated as part of the team early on, a data scientist will learn how to support the team in an effective way. When I started in my role in health care, I was fortunate to have a nursing leader in critical care allow me to shadow her for my first week. I learned more in that first week about the health care culture than in the next 5 years I spent in that role. I would also tell clinical leaders that a data scientist should be a part of the process every step of the way, not just at the beginning or end of an intervention/initiative. Ask the data scientist what they can do for you from start to finish, and take them along on the journey as a part of the team. They will thank you, and you will be grateful in the end!

Share this article
Subscribe to Blog