The Challenge of Unusual Events and Big Data in Healthcare

By Alan Spiro, SVP & CMO, Blue Health

The troubling story in The Wall Street Journal describing “… the fight for Whitney Brown” is an account that brings to life some of the problems of our system today. In reading the article, one cannot help but wonder if faster action could have saved Ms. Brown. If it were not raining and a helicopter could have made it through, might that have been enough to save her? If the EMS-rescue service had been willing to bend the rules and allow more ambulances out of the area to save time, could that have forestalled the cardiac arrest that led to brain damage? There is no way to know. As noted in the article, amniotic fluid embolus is a rare but devastating condition that often kills – even when everything is done quickly and correctly. Here, it appears that everything was done correctly but not quickly.

The practice of medicine is often geared to the identification of those unusual and devastating events in a timely enough manner to intervene and save the patient. When I was a practicing gastroenterologist, a primary care physician asked me to perform an endoscopy to try and diagnose the source of a patient’s severe abdominal pain. But the endoscopy did not help find a diagnosis, and it turned out the patient had kidney cancer – something totally unrelated to the GI tract that I was investigating. Fortunately, in that case – when the clue of microscopic blood in the urine was pursued – the correct diagnosis was made quickly enough to impact the course of the disease.

In my medical training, we were taught that the expert physician could evaluate all the signs, symptoms and data together and develop a comprehensive list of what could be causing the problem – no matter how unusual the diagnoses were toward the bottom of the list. Success was defined by timely diagnosis and treatment – even if the patient suffered from an illness never seen previously by you, the physician.

In all my years of Gastroenterology, for instance, I never saw a person with a case of Chagas Disease. But I always considered it as a diagnosis. If you don’t think about the unusual (and you often don’t need to actively look for rare illnesses because the problem turns out to be something higher on the probability list), you have not done your job as a thoughtful physician.

But in an era of big data in , this core concern of medicine – that of identifying and treating the unusual event before it can become devastating – can be minimized. Unusual events in large databases are often viewed as “noise” in the graph and purposely ignored to make sense of the data. If you are looking for trends in populations to set policy and develop plans for populations, you can miss the unique problems and needs of the individual. If you only employ data to effectively and efficiently treat those with common diseases, the unique problems of the individual can be lost.

Most people are unique. As a consultant to large organizations that employ data to find ways to create benefits and programs for their people, I was often asked to identify the category of diagnoses that was responsible for the largest portion of their costs. The clients’ guess was usually musculoskeletal disease or cardiac disease. But I would then show them the data, which typically revealed the largest category to be “other.” In medicine, in aggregate, the unusual is often usual.

As a leader of a company that has one of the largest repositories of data – comprised of more than 172 million lives – I understand the challenge is often one of not allowing individuals to get lost in the tidal wave of big data. Our challenge is to find better ways of identifying people like Whitney Brown before the catastrophe occurs so you can introduce measures to proactively treat problems. It often means combining different sources of data and determining – through statistics, predictive and mathematics – if different approaches are warranted. Specifically, should a young woman with two miscarriages and a history of opioid addiction – who lives in a high-risk pregnancy care desert – be housed, instead, in a hotel close to a perinatology center in her final weeks of pregnancy? Our challenge is to use big data in creative new ways that look beyond the broad trends to impact those individuals who are cursed by being at risk for the unusual devastating event that every physician is trained to look for. I am proud to be with an organization that is dedicated to doing exactly that.

Leave a Reply

Your email address will not be published.