Women of Childbearing Potential and Data Science in Action

June 11, 2018

In clinic recently, I saw a 35-year-old woman who was suffering greatly from rheumatoid arthritis (RA), an autoimmune disease that causes chronic, painful inflammation and sometimes fusion of multiple joints. There are more than a dozen FDA-approved treatments for RA, and “Jane” had been taking one with good results. But two years ago when she began trying to get pregnant, her rheumatologist and her obstetrician advised her to stop taking her medication due to worries about possible effects on a developing fetus.

Over the next two years, not only had Jane not gotten pregnant, but she was experiencing significant pain. Her wrists were now severely swollen with very limited movement and likely had been permanently damaged. She had two questions for me: did she have to live in pain, and how could she get pregnant? 

As a rheumatologist with a focus on pregnancy, I knew that there were ample data about the safety of the drug that had worked so well for Jane—data that showed no increases in birth defects, pregnancy loss, preterm birth, or neonatal infections. I told Jane she could resume taking her medication, and she and her husband left happy at the prospect of relief. I expect to see her back pregnant soon – we know that active RA prevents pregnancy in many women.

This time, I was able offer good news. But there is a general shortage of information about the safety of medications used during pregnancy—largely because any woman who is pregnant, was recently pregnant, or might get pregnant is barred from participating in most of the clinical trials that evaluate drug safety and efficacy. The upshot is that many drugs are approved by the FDA with little or no data about how they affect pregnancy or offspring.

In some cases, FDA may require manufacturers to establish a pregnancy registry for a new drug. But while these registries can provide detailed information, they are very slow to enroll. For examples, the registries for rheumatoid arthritis therapies approved by the FDA in 1999 have taken over 15 years to obtain useful data. A recent draft guidance on the inclusion of pregnant women in clinical trials issued by FDA is a step in the right direction, but more still needs to be done.

Ironically, doctors and their patients are suffering from a lack of actionable data while at the same time the overall pool of potential information is growing rapidly. If we can find ways to leverage “big data” resources and apply the right kinds of analytical tools, we can close these gaps in knowledge and ensure that pregnant women have access to high-quality data about the safety and effectiveness of medical products for them and their offspring. Here are several ways this could happen:

  1. Locate new pregnancies for prospective registries. Large datasets from hospitals, payers, and other sources can be used for real-time identification of new pregnancies in women who are likely taking medications of interest. These women can then be invited to participate in prospective registries, removing a key roadblock to registry accrual while also improving data accuracy.
  2. Data exploration: Large datasets could be mined to connect prescription, medication use, pregnancy outcomes, offspring health, and to identify safety signals in large populations. 
  3. Improvements to electronic health records (EHRs): EHRs should be updated to demand accurate medication exposure data during pregnancy. Obstetricians and other clinicians could enter this data, which would then be confirmed by the patient – an essential step to ensure accuracy. Such data could fuel prospective cohort studies of pregnancies and offspring after exposure, as well as case-control studies of birth defects by providing accurate, real-time data about maternal exposure.

But although technical solutions are within our grasp, lawmakers, pharmaceutical companies, and regulators must first take action. Requiring close monitoring of all pregnancies conceived during a drug trial and reporting accurate pregnancy outcomes would be a great start. When a women in a clinical trial falls pregnant, she could simply stay on trial without taking study medication, or enter a separate pregnancy cohort study—and my own experience suggests that many pregnant women would be willing and even eager to contribute to needed medical knowledge.

Studying pregnancies may seem risky, but it is essential to our ability to improve the health of women and babies. When we realize that untreated diseases often pose a greater risk to a pregnancy and offspring than do the treatments for them, it’s unethical to refuse to study the impact of medications and disease control on pregnancy and offspring. And while we can do more to capitalize on currently available resources, we must also build new tools and approaches that facilitate collection of accurate, reliable data. By doing so, we can take the guesswork out of treating pregnant women—and give them access to the evidence-based care they deserve.

See more blog posts for Dr. Clowse


Portrait of Duke Forge Co-Director Megan E.B. Clowse, MD. MPH