Back in the summer of 2010 when Jenny was a second-year resident in Internal Medicine, she called me out of the blue to say she didn’t like the way we were taking care of patients with Staphylococcus aureus bloodstream infection. “We should be making better antibiotic choices from the beginning. We should be evaluating their heart for evidence of endocarditis,” she announced. I agreed with her. “Do you want to work on this? A lot of times, we know the gaps in care. We don’t have the ‘bandwidth’ to work on all the gaps at the same time.” She said she is very passionate about the issue and that she will work on it during off-hours and weekends. I told her we can set this up as a Quality Improvement project or a hypothesis-based research study. I wasn’t really sure at that moment what the hypothesis would be (as the guidelines were already there) and resisted the urge (which is common among faculty in competitive academic environments) to pontificate that it takes several years for published guidelines to translate into routine clinical practice. She told me all she wanted was to see the care improved. She was clear she didn’t care about getting a publication out of this work. The picture above is the article that’s just published, 5-1/2 years later. The complete article is available at:

In spite of our busy schedules, we made the time to meet frequently to go over writing the protocol, the application for the Institutional Review Board, setting up the data collection tool in an Excel spreadsheet and a dictionary of terms we were going to use for data collection and their definitions (the ‘variable dictionary’). By the time she collected data on the 167 patients with bloodstream infection caused by Staphylococcus aureus during the ‘baseline period’ of one year, she graduated from residency and began her fellowship training in Infectious Diseases, thankfully in our ID division. As she had already spent four years of medical school and three years of residency in Internal Medicine in the UT Southwestern and Parkland environment, she knew the key players well. Putting together a list of ‘what needs to happen’ was easy. Getting a consensus among the Infectious Diseases faculty members and the Cardiology faculty members was a little more effort. Teaching the guidelines to the residents and use those in practice took some more effort. In academic medical centers, it’s the resident trainees who write treatment orders for patients. Even though they work under attending supervision, not all attendings ‘micromanage’. There is graded autonomy and there is a time crunch to round on a large number of patients in a limited time. In addition, everyone works in 2-4 week blocks of time on any particular team. Jenny made educational pocket cards and gave them to residents in internal medicine, emergency medicine, and general surgery, and also enlisted the chief residents to make sure they were handed down generations of residents. Just so it’s easy to view all the recommendations in one place, we created a treatment ‘order set’ in the electronic medical record so that the residents can check or uncheck orders instead of having to remember writing each one of them. We measured the clinical characteristics and clinical outcomes of patients with Staphylococcus aureus bloodstream infection before and after the intervention.

Comparing the 127 infection cases after the intervention with the 167 infection cases before the intervention was implemented, more patients got better care after the intervention. Almost half the cases (44%) received a transesophageal echocardiogram after the intervention compared to only a fifth of them (20%) before the intervention. Time to start appropriate antibiotic therapy improved from 5 hours before intervention to less than an hour after intervention. More patients in the post-intervention period received consultation by the Infectious Diseases service and referrals to follow up in the Infectious Diseases clinic after discharge from the hospital. These infections are notorious for relapsing after completion of treatment and a real risk of death. The usual rates of death among these patients are 20-40%. Even before intervention, the death rate in our Parkland patients at the end of 90 days from the date of infection was low at 12.5% and the rate of relapse was 11.9%. They reduced to 12.2% and 2.4% respectively after the intervention was implemented. A key lesson learned was that although we spent a lot of time building the order set in the electronic medical record, the residents did not use the order set to write orders. At the time they found themselves writing treatment orders on patients with Staphylococcus aureus bacteremia, they did not ‘remember’ to use the order set. In future, we need to find ways to ‘prompt’ physicians to use these order sets.

Such quality improvement projects are routine in healthcare systems. What’s special about this project is that one of our trainees initiated the project and carried it through completion. There was no formal project management team for this effort. My role as mentor was to support her, set up proper structures around the project, facilitate the implementation and troubleshoot as needed. The most critical factor was the engagement of residents in improving quality of care by someone who is also a resident trainee herself. This intervention was successful even in an environment where the hospital is a busy resource-limited setting and the residency program is extremely intense and competitive. This should be possible in other clinical settings as well.