Dr. Andrew Lee and Dr. Andrew Carey appear on another episode of NeuroOp Guru to discuss evidence-based assessment of optic neuritis through big data analytics.
Video Transcript
Editor’s Note -This transcript has been edited for clarity.
Andy Lee, MD:
Hello and welcome to the latest edition of NeuroOp Guru. I’m here with my good friend, Dr. Drew Carey from Johns Hopkins University. Hi, Drew.
Drew Carey, MD:
Hello, this is Dr. Andy Lee.
Andy Lee, MD:
Today we will talk about evidence-based assessment of optic neuritis through big data analytics. Dr. Carey, what is big data?
Drew Carey, MD:
Big data is a term that refers to very large numbers that typically come from health systems, insurance databases, or Medicare-based populations. They are too large to sift through individual patient charts and require the help of a statistician or other type of evaluation to pull the data from what is pre-programmed into the system of record.
Andy Lee, MD:
And what significance did this have for the research methodology?
Drew Carey, MD:
The study queried 6.7 million medical records.
Andy Lee, MD:
It’s quite big.
Drew Carey, MD:
It’s quite big, yes.
Andy Lee, MD:
What were the results? Can you explain these two tables in more detail?
Drew Carey, MD:
Yes, that’s right. So what we did was we queried the dataset for optic neuritis diagnoses, neurological and systemic inflammatory diseases. And then we looked at the overlaps to see what associations existed. From 6.7 million charts, we identified 5,300 cases of optic neuritis, and of those, about 52% were associated with some underlying etiology that we could identify or link to. The most common was multiple sclerosis, which accounted for 21% of the optic neuritis cases. This is not surprising, I think, because we found in the ONTT data that about 50% of patients were associated with multiple sclerosis. Of course, this is a prospective study, so we’re able to follow patients over time, and even if they didn’t meet the diagnostic criteria for MS when they first developed optic neuritis, they could develop it later.
The second most common was neuromyelitis optica spectrum disorder, which accounted for about 11% of the optic neuritis cases identified in this study. The third most common was gout, which was present in 4.2% of the patients. I don’t think the authors or anyone else would think that gout could cause optic neuritis, but gout is very common. Sjögren’s syndrome [and] Sarcoids are both very common in the 3-4% range, and less common ones like rheumatoid arthritis, HLA-B27 associated disease, lupus, and mixed connective tissue disease are in the 1-2% range, while scleroderma, MOGAD, GPA, granulomatosis with polyangiitis, and Behcet’s disease are all less than 1%.
However, this requires a bit of caution. This covers about 10 years of data, and there was no good test for MOGAD throughout that time. So there may have been a lot of patients who were not diagnosed during that time because there was no test for MOGAD. So the authors did a subset analysis of patients with these high-risk diseases, and they looked at a two-year period to see which diseases were at the highest risk for optic neuritis. In that respect, patients with neuromyelitis optica spectrum disorder, who were known to have optic neuritis, had the highest incidence of optic neuritis, with a relative risk of about 1,200 compared to the general population. The next most common was MOGAD, with a relative risk of about 900 compared to the general population. And then there’s MS, with a relative risk of about 125. It’s still a very high-risk disease, but it’s sarcoidosis, Behcet’s disease, Sjogren’s syndrome, related diseases, and less common diseases like lupus and mixed connective tissue disease.
Andy Lee, MD:
So, what was the conclusion of this big data analysis?
Drew Carey, MD:
Yes. Optic neuritis was fairly common in the population. There were about 8 cases per 10,000 medical records that existed. This is consistent with other studies, and I would say that on average it’s about 1 case per 10,000 population. Of course not. If you look at the medical records, these patients are seeking medical care. About 48% of the patients were idiopathic or no findings were found, and no underlying cause was identified. Multiple sclerosis was the most common, followed by neuromyelitis optica. Neuromyelitis optica and MOGAD had the highest incidence of optic neuritis, followed by MS. The bottom line is that when you see a patient with something like optic neuritis, you should definitely think about the demyelinating diseases MS, NMOSD, and MOGAD, and you don’t necessarily have to test for them, but at least take a history and do an evaluation with a test to see if you should think about those diseases. Diseases like sarcoid, Behcet’s, and Sjogren’s syndrome are other diseases to keep in the back of your mind and worry about in patients who fit the clinical picture, or who don’t look like that on an MRI, or who don’t have other clinical features that would suggest a primary demyelinating disease.
Andy Lee, MD:
Is there a message you want to convey to your audience? Will we see more big data in the future? Do you think it will be useful? How will it be useful?
Drew Carey, MD:
Yeah, I think we’re definitely seeing an increase in big data research. I think big data can help us put together evidence-based differential diagnoses. We want to think that common things are common, so if a patient doesn’t have one of the common symptoms that we see every day, that can help us prescribe a workup. And if we can move to a value-based care and cost-saving approach, maybe we can lean the health system a little bit by approaching differential diagnoses in a risk-oriented way. And as electronic health systems improve, maybe we can do reflex tests instead of shotgun tests, which might help save money on health care. And as big data gets more and more sophisticated, maybe we’ll be able to put patient information into a risk calculator and think about what’s most concerning to leave in the differential diagnosis. Or maybe a physician assistant, an AI guide tool, will use the chart to prescribe eye drops based on the patient’s medical history, that they have a dry mouth, that they’re using three different eye drops for dry eyes, that they have optic neuritis, etc. Maybe we need to think about Sjogren’s syndrome as an underlying etiology. Big data allows us to do analyses that we can’t do manually, to find associations that we can’t do in small series. This increases the diversity of patient samples and provides more generalizable information.
Andy Lee, MD:
Dr. Carey, that was a really interesting talk. Thank you so much for your time. This concludes the latest edition of NeuroOp Guru.
Drew Carey, MD:
Thank you for inviting me.
