Summary by: Arooba Ahmed (CC '23)
Dr. Hripcsak talks about Observational Health Data Sciences and Informatics (OHDSI), an International network of patient databases of which Columbia is the coordinating institution. Currently, there is not much COVID information, but there are 350 researchers working to see what information can be extracted.
Dr. Hripcsak shows what they were able to do in the first few days, beginning with the observation that the age distribution for social groups who were infected early in Korea was between 20-44 years. At Columbia, the age distribution was between 60-80 years. These distributions show an infection of young people in a disease of elderly people. The Korean population gives a sense of the sub-sample the Columbia researchers are using, and could imply that when they were taking measurements here, they were seeing just a small sample of the total infected population in NY. Of the raw data at Columbia, they also saw that there was a bias, where the people coming and warranting tests also have hypertensive disease diagnosis.
He also shows two studies regarding Angiotensin receptor blockers. The study wanted to see whether people who have chronically been taking them are in worse condition. They tested ACE/ARB use in hypertensive patients to see how they do. They do not have any statistically significant information yet for the Korean patients, but they are coming up with new ways to test the causal model. However, at Columbia, they do not see a major detriment for being on ACE or ARB. These differences in studies are important because when you go online and read papers for simple observational studies and the covariance in this disease, you have to be very careful to interpret it. Simple correction factors on the data can drastically change the outcome.
The best way to study these drugs is through randomized trials, but some observational data is included in Dr. Hripcak’s presentation. The FDA and the European FDA (EMA) have asked for the safety profile of the drugs (such as hydroxychloroquine ), so they did a study of important outcomes on 130,000 patients who took hydroxychloroquine versus a comparative drug. They did not see any major side effects. This is the first study of the international size which has a very small confidence interval for hydroxychloroquine, because even though it is an old drug, its safety has not really been studied.
They are also setting up a Columbia COVID Data mart in the tri institutional data house, which is called Jupiter. They gather demographic and lab data (among other things) such as procedures and vitals. Because they also have limited data right now, they are conducting manual data extraction projects.