'Long Data': A Missing Link in Patient Histories


October 10, 2018

We've heard a lot of talk and buzz about "big data" in recent years, but what we're beginning to learn is that we're missing the "long data": longitudinal data from the prenatal period all the way through our lifespan.

We were awakened to this potential deficit by a recent study of adolescents conceived by in vitro fertilization (IVF).[1] Recall that we're in our 40th year of IVF and more than 6 million people have come into the world via this route. Everything seemed to be perfectly normal about IVF-conceived individuals, and that may indeed still be the case. But a study by Meister and colleagues from the University of Bern in Switzerland raised the concern that IVF is associated with premature vascular aging.

They compared 54 teenagers conceived by IVF with 43 age- and gender-matched controls. All of the vascular parameters that were measured suggested abnormalities in the former group, including flow-mediated vasodilation, pulse-wave velocity, carotid artery intima-media thickness (Figure), and 24-hour ambulatory blood pressure. Of note, these same youngsters had participated in the study 5 years before with a similar battery of vascular tests, and the same trends were evident, albeit not as significantly.

Figure. Premature vascular aging in IVF children persists into adulthood.[1]

(A) Flow-mediated dilation (FMD), (B) pulse-wave velocity (PWV), and (C) intima-media thickness (IMT) at baseline (mean ± SD age, control 11.8 ± 2.2 years; assisted reproductive technologies [ART] 10.9 ± 2.4 years) and at 5-year follow-up (mean age; control 17.6 ± 2.7 years; ART 16.6 ± 2.4 year) in participants conceived through ART and naturally conceived control participants. Premature vascular aging, as evidenced by impaired flow-mediated dilation and by increased PWV and IMT, persists at 5-year follow-up in participants conceived through ART. Data are shown as mean ± SD.
Republished with permission from the Journal of the American College of Cardiology

The findings of premature vascular aging have been noted previously in mice that were conceived via IVF compared with controls.[2] A theory to explain this invokes the epigenetic changes of the embryo related to IVF, which may be particularly susceptible to early cardiovascular development. But we don't know whether this is real. It would require independent replication in much larger samples of people.

The point, however, is that we could have missed this whole story because we don't collect long data.

When conducting a medical history, I have never asked a patient, "Were you an IVF baby? If so, was the embryo transfer fresh or frozen?" Nor do we ask women whether they had gestational diabetes when they were pregnant, which is a recognized risk factor for developing type 2 diabetes later in life. We don't ask patients if they were born prematurely, how much before term, and what, if any, complications were manifest. But these early-life factors could turn out to be quite important decades later in life.

Beyond these concerns, too many decisions in medicine today are not made with full knowledge of the patient's prior history. While much attention has been paid to medication allergies, so many other factors are overlooked when tests are ordered or surgeries performed. I had firsthand experience when undergoing a knee replacement a couple of years ago. What had brought me to that need—a rare bone disease that I had as a kid (osteochondritis dissecans)—was not considered when the orthopedist decided on the type of surgery or the optimal post-op physical therapy.

Ilana Yurkiewicz, MD, of Stanford University, recently wrote an extraordinary essay: "Paper Trails: Living and Dying With Fragmented Medical Records." In it she takes us through the experience of a patient for whom seriously flawed care was rendered due to incomplete access to medical data. One section sums up the problem so well:

It's like opening a book to page 200 and being asked to write page 201. That can be challenging enough. But on top of that, maybe the middle is mysteriously ripped out, pages 75 to 95 are shuffled, and several chapters don't even seem to be part of the same story.

As we move forward in the world of artificial intelligence with the capability of deep learning—an input and output story—for each individual, the ability to accurately predict risk or provide meaningful guidance is going to be predicated on having comprehensive data. There is already too much data for clinicians to handle for patients who have many healthcare providers, lab tests, scans, medical encounters, medications, and social history. Adding to that are the roughly 20 million Americans who have had a consumer genome-wide scan.

Also, wearable medical sensor data for continuous heart rhythm or glucose levels are being used more and more, further expanding the tally of big data per individual. Thus, the notion of collecting even more data may seem unattractive.

But as we consider the trend of "data-fying" medicine, we should not discount or forget how important long data may turn out to be. Who knows how much we've missed over the years by not routinely capturing data from our point of conception?


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: