Let's talk about Surveillance Bias and Public Reporting. People ask me all the time, how did I get interested in VTE or venous thromboembolism? So, it started when I was the Performance Improvement director for the Adult Trauma Center at Johns Hopkins Hospital. We're a level one trauma center, we're accredited by the state of Maryland. And we got a letter which basically said, "Dear Johns Hopkins Adult Trauma Center, You have the highest DVT rate of all the Trauma Centers all across Maryland. Why are you such bad doctors? Sincerely, MIEMSS" which is our accrediting organization. Needless to say, my boss and I were not very happy to get this letter and it started us thinking. And this is really where this research idea started. We knew that we were very aggressive about screening and looking for DVT in high risk asymptomatic trauma patients. And we started to wonder, what do other trauma centers do? Are other centers doing the same as us or we're different? And does this impact the reported rates of DVT? Maybe this is the reason. So, there's conflict out there about the use of duplex ultrasound screening for asymptomatic trauma patients for DVT. And the data looks something like this, the pro side is that if you screen patients who are asymptomatic, find a DVT, you can find the DVT early, and allow treatment with anticoagulation before the DVT breaks off, and becomes a fatal PE, and the patient dies. So, that seems like a good strong pro. However, there is a con side. It can be quite expensive to do ultrasounds in every single high-risk trauma patient. It may not be cost effective and probably more importantly, there might be harm from anticoagulation. Say you find a small DVT in the lower leg that would never have caused a problem, would never have caused symptoms, and you anticoagulate the patient, and they bleed from their anticoagulation. You might have done them harm and that might be a drawback to screening. So screening is not free whether it be on the cost side or from the patient's perspective. There's actually conflicting guidelines on the topic. So, there are two large national organizations that have very different opinions on whether we should be screening high-risk asymptomatic patients for DVT. So, EAST and CHEST, or the American College of Chest Physicians, have very different opinions based on the same data. EAST suggests that, "Serial duplex ultrasound imaging of high-risk asymptomatic trauma patients may be cost-effective and may decrease the incidence of PE." That seems like a good pro side that you might do it. However, the American College of Chest Physicians Guideline says that, "For major trauma patients, we wreck command or we suggest that periodic surveillance with venous compression ultrasound or duplex should not be performed." So now, what's a clinical provider to do when you're getting differing opinions from two important organizations? Well, we started to look at some data and we started at our hospital. So, this is a single center study from Johns Hopkins Hospital, and this is a before versus after screening after a guideline was implemented. And you can see that in the before period, we were doing some ultrasounds in finding some DVT, but in the after period, we did significantly more duplex ultrasounds per patient, and found a tenfold higher rate of DVT in these patients as well. We're looking and we're finding more. So then, we wanted to know, is this a problem just at our trauma center or is this something that happens on a national level? So, we went to the National Trauma Data Bank which is the largest aggregation of trauma data in the world. It's run by the American College of Surgeons. And I had a mentee who did his master's thesis on this exact question and here's the data, when you break out hospitals by quartile based on how many ultrasounds they're doing per patient, you will see that the hospitals in the highest quartile of duplex ultrasound have a sevenfold higher rate of DVT than hospitals in the other three quartiles. It seems as though the hospitals that are doing more ultrasounds are finding more DVT. Now, the arguments we got about this data was this is at a hospital level, maybe those hospitals really are seeing really sick patients, and that's why their DVT rate is higher, and they're doing ultrasounds because the patients are sick, and have DVT. That is a certainly a reasonable question, so we went ahead and did another study. So, we published this in The Journal Trauma and we showed that hospital screening status is actually an independent risk factor for DVT reporting. We know there are many risk factors in trauma patients for VTE. And here are some of them listed: age, extremity injury, head injury, ventilator days, venous injury, major surgery, all those things are associated with DVT in trauma patients. But even when we control for all of those things on a patient level, what we find is that patients seen at a "screening" trauma center, meaning one that does more ultrasounds rather than a "non-screening" trauma center based on doing fewer ultrasounds, those patients at the screening centers are twice as likely to have a DVT identified, and put into the trauma registry. I'm not saying they're twice as likely to have a DVT in their leg but that screening identifies the DVT, and gets it into this registry, and that's where this reporting issue comes about. There's still variation out there about what do trauma providers think we should be doing. So, when asked a question of, "Should high risk asymptomatic patients be screened for DVT?" About three quarters of trauma surgeons surveyed, agreed. But what that means is the other quarter or so did not agree or disagreed or didn't really know what to do, so they didn't really have a preference. So, there's variation out there in what trauma surgeons are doing. And this is a classic example of surveillance bias. Providers who screen more aggressively by performing more duplex ultrasounds may identify more cases of DVT and appear, the keyword is appear, to be providing worse quality of care than those who order fewer tests. So,what are the implications of this? We know that this variability in DVT screening likely leads to the variability in DVT rates reported and that these DVT rates are biased. They are not truly indicative of the care we provide. And we published this concept in JAMA with my quality and safety mentor, Peter Pronovost.
