Quality of Data
Hello, I’m Dr. Jeff Kingsley and welcome to another edition of Riding in Cars With Researchers. Today I want to talk to research coordinators and physicians. I want to talk about the quality of your data, how to think through the data points that you’re collecting in research trials.
In normal healthcare, we teach young doctors, residents, and medical students to treat the patient, not the data point. We teach them that it’s the constellation of data points that means far more than any individual data point. Let me give you some examples of that. ST segment elevation in one lead of a 12 lead EKG….what does that mean? What is that? ST segment elevation in three contiguous leads that are meaningful for damage in a specific area of the heart. That means something far more. ST segment elevation in three contiguous leads with a patient who has acute substernal crushing chest pain, diaphoresis, nausea, shortness of breath, and radiation of the pain to the jaw and left arm.
The constellation of symptoms and signs means far more than any individual one of those signs and symptoms independently. Same thing happens in research, but for some reason in research, everyone jumps to believe any individual data point as being meaningful or telling examples. I had a patient last week who is doing well in a NASH trial (non-alcoholic steatohepatitis). We’ve been doing routine labs and everything has been fine. We get her routine labs and she has stark elevations in liver enzymes. Everyone becomes immediately concerned and my email blows up immediately to stop the study drug and do an early termination visit on this patient. Well, wait a minute, this doesn’t make sense. Nothing has changed and she’s asymptomatic. We’ve had other labs that were all completely normal, so why, all of a sudden, can these labs be so abnormal…..it doesn’t make sense. Let’s immediately repeat the lab. So we repeat the lab and the lab results are completely normal.
Treat the Patient, Not the Data
Treat the patient, not the data point. If you do this, you will have better quality data in your research trials. You will have better quality adverse event profiles. The adverse events that you document on your patients will be far more likely to be truthful and accurate then if you do otherwise. The patients that you do early terminate, or patients that you screen fail, will be far more appropriate to have been early terminated in a trial or to screen fail. We need to treat these data points differently in research than we do today. We need to treat them more like we do in normal healthcare.
Last example, I had a patient who screened yesterday for a pilot nephritis trial. A patient was seen in the ER, diagnosed with pyelonephritis, a kidney infection, and was screening for a pilot nephritis trial that we have going on. The patient’s liver enzymes were elevated and the Sponsor said to immediately screen fail the patient because the liver enzymes were elevated. This is a 19 year old girl with elevated liver enzymes. Why does she have elevated liver enzymes? We immediately repeat the labs. We repeat the labs and, low and behold, the liver enzymes are not only elevated, they are more elevated including alkaline phosphatase. It’s likely that the patient has the wrong diagnosis. It’s a 19 year old woman who has a UTI and because of her abdominal and back pain, the diagnosis of pyelonephritis was made. It’s likely that this patient needs an ultrasound of her biliary tree and actually has two concurrent diagnosis. Patient is screen failed, but we served the patient much better by not simply screen failing the patient, but by digging deeper and looking for a constellation of symptoms.
Send me topics you want me to talk about at firstname.lastname@example.org!
Subscribe to our blog HERE