2013 Vision Grant Recipient Provides Update

2015 - 2 April – Clinical Controversies
In this article, the 2013 Vision Grant recipient provides an update on her grant-funded research.


Support from the Society of Critical Care Medicine’s (SCCM) 2013 Vision Grant enabled Azra Bihorac, MD, MS, FCCM, to conduct research centered on developing computer algorithms that use existing clinical data to efficiently and accurately predict the risk for acute kidney injury (AKI) in surgical patients. Dr. Bihorac, an associate professor of anesthesiology, surgery and medicine at the University of Florida and senior intensivist and nephrologist at University of Florida Health, took time out of her busy schedule to answer some questions related to this research.

What led to your interest in this specific area of research?

My research career started in the mid-1990s during my subspecialty training in nephrology in Istanbul, Turkey, where my family and I fled after war erupted in my home country of Bosnia and Herzegovina. At that time my research involved hypertension and cardiovascular disease in patients with chronic kidney disease, but I was fortunate to be exposed to the multidisciplinary concept of critical care medicine as I trained in the first medical intensive care unit (ICU) in Turkey.  It was also in Turkey that I witnessed AKI in epidemic proportion after the tragic earthquake in 1999 when I was involved in organizing renal support for multiple victims with crush injuries and AKI. After moving to the United States in 1999 as a fellow of the International Society of Nephrology, I was determined to continue a research career centered on better understanding the mechanisms of kidney disease in patients with hypertension. My clinical experience, however, reshaped my research curiosity. Although patients with AKI constituted the largest proportion of patients in our in-hospital nephrology practice, it was quite frustrating that we almost inevitably would arrive too late to offer anything more than support with dialysis. I wanted to be a first—not a last—responder for patients with AKI. In other words, I wanted to make a difference by preventing AKI, rather than supporting the completely failed kidney. The ICU was a natural battlefield for AKI; I saw intensivists as hospital first responders whose decisions shaped not only immediate but also long-term health outcomes for critically ill patients. After completing training in surgical critical care, I started my academic and research career in the division of critical care medicine, taking care of surgical patients in the ICU.  My research goals are focused on the development of clinical tools that can help surgeons, anesthesiologists and intensivists prevent AKI whenever possible, mitigate the progression to kidney failure once injury has occurred and facilitate recovery of kidney function when that is possible.

Can you describe your Vision Grant-affiliated research project?

In our previous research, we demonstrated that AKI is one of the most common and devastating—yet poorly understood and recognized—ICU complications in surgical patients. The main objective of our Vision Grant project was to develop computer algorithms that can use existing clinical data to efficiently and accurately predict the risk for AKI in surgical patients. Our hope was to eventually use the algorithm to automate the process of clinical risk stratification in the clinical workflow when incorporated into electronic health records. We began with describing steps for data normalization and then developed mathematical approaches to capture the complex association between changes over time in clinical measurements, such as blood pressure, heart rate and creatinine, with 90-day mortality after surgery. As a final result, we have internally validated two predictive models that can stratify patients’ risk for AKI both preoperatively and in the immediate postoperative period and which will allow physicians to initiate AKI preventative measures at ICU admission. (Both algorithms are developed as R packages and will accompany the peer-reviewed articles in preparation.) 

What was the most surprising or challenging aspect of your research?

The amount of factors at play. There are many factors we often do not think about when using our clinical judgment for AKI risk stratification that played an important role in developing our computer algorithms. For example, the importance of a patient’s residing address or a patient’s operating surgeon in the algorithm was quite surprising. The fact that certain associations may emerge that are difficult to explain using our conventional thinking is a challenge in research that uses big datasets and computer algorithms. We will need to become more adept at interpreting and judging the quality of such research, as was recently highlighted in the TRIPOD (Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis) Statement.

What are your future research plans?

Using the results and experience from the Vision Grant research, we plan to apply for NIH (National Institutes of Health) funding to perform prospective validation of a clinical decision support tool for AKI risk stratification in the perioperative period, and to improve the process of identifying patients who need further testing with AKI biomarkers. This two-step approach will hopefully further increase the predictive accuracy of risk stratification for AKI after surgery.

Do you have any advice for future Vision Grant applicants?

Be persistent and use the reviewers’ feedback to improve your grant and resubmit!

Do you have anything else you would like to share?

The Society has played a great role in my growth as an intensivist. It has been a place for networking and a resource for knowledge and research ideas. The Vision Grant provided me with resources to assemble my team and get my research off the ground. I encourage all my colleagues to become involved in Society activities. By offering up their time and expertise, they will help our profession meet the growing challenges in healthcare.