Last time (link), we started talking about ways to visualize Surgical Infection data. The dashboards from last time show hospital-wide performance, which is definitely a big deal. When I first presented this information to hospital and physician leadership, one of the main questions they had was how individual doctors compared to each other. This seemed like a great opportunity to demonstrate the power and flexibility of the SIR!
Reminder/ICYMI: the SIR is the Standardized Infection Ratio. It compares the observed number of infections to a risk-adjusted number of predicted infections. SIR = sum(observed infections) / sum(predicted infections). This allows us to compare surgeries with different risk levels to each other on an equitable basis. If your SIR is greater than 1, you aren’t doing so well, but if it’s less than or equal to 1, great job!
Our data source gives the risk (i.e. predicted number of infections) for each individual surgery, so we can slice the data however we want. Please note that all data shown below is faked due to confidentiality issues.
Start with a scatter plot
I started by creating a scatter plot comparing SIR to number of procedures done, with each dot representing an individual surgeon. I put a reference line at 1.0, to better indicate whether a surgeon had more or less infections than expected. Click here to see the interactive version of this dashboard.
One of the first things we noticed was that there seemed to be a link between more procedures done and a lower SIR (and vice versa). This agrees with what we might expect, but it’s helpful to be able to show this with data. Keep in mind – anyone who didn’t do enough procedures to have at least 1.0 predicted infections is excluded from the scatter plot above.
Next, we wanted to have a way to dig deeper into this data. I created a sorted bar chart showing SIR by procedure type, and added this to a dashboard with the scatter plot. By using Action Filters, I was able to introduce some interactivity, letting users start to answer their own questions.
Let’s add some trends!
The next question that presented itself was whether performance was getting better, getting worse, or staying the same. To answer this, I added a graph showing observed vs predicted infections over time. Again, I used Action Filters to introduce interactivity.
In the example illustrated above, we drill down to look at Abdominal Hysterectomies, and then look at the surgeon with the worst SIR. The trended data shows that they were doing poorly, but have improved considerably in the last few quarters.
This level of insight and interactivity has proven immensely valuable to our department. Our Hospital Epidemiologist has been able to gather best practices from the highest performing surgeons and work closely with the surgeons with the highest SIRs to improve their practice. While this technically would have been possible before, this dashboard allowed us to quantitatively show the low-performing surgeons that they needed to improve.
Thanks for reading! Click here to see the interactive version of this dashboard. Stay tuned – we’ll keep looking at ways to visualize Healthcare Quality and Infection Prevention data in the weeks to come.