Millions of pregnant women and future generations depend on our ability to understand the impact of Covid-19 on pregnancy outcomes. Ob-Gyn physicians are coming together in informal groups on Facebook and discussing anecdotal up-ticks in preterm birth, stillbirth, and pregnancy losses. Unfortunately, there is no comprehensive pregnancy data set available to inform national guidelines and validate or refute these anecdotal observations. This problem is worth solving.
There are many reasons these highly valuable data sets do not exist, ranging from technology limitations to insufficient voluntary provider and patient data reporting to a lack of clinical data aggregation.
Electronic Health Records (EHRs), the underbellies for hospital and ambulatory practices across the country excel at aiding Ob-Gyn practices in routing claims, connecting to labs and pharmacies, and supporting revenue cycle management. However, these platforms do not capture and discretely store valuable clinical data, but rather store it in narrative notes or pre-defined (and commonly cloned) templates. In the FDA’s Framework for Evaluating RWD/RWE the agency states,
“EHRs and medical claims data may not capture all data elements needed to answer the question of interest. We expect that these sources will generally record major events like hospitalization, but other changes in medical status (e.g., worsening of depression or anxiety, increased joint pain, changes in severity of a dermatologic condition, worsening asthma) may not be reliably and consistently documented in the EHR, if at all. Even when captured, the way the data elements are captured in the EHR may limit their accessibility.”
Solutions are desperately needed that maintain these critical patient data elements in longitudinal discrete data stores and normalize across geographically distributed practices. When solutions organize data discretely, the data underlying real-world clinical questions is more readily accessible for a quick and thorough review. Guidelines aimed at protecting and saving lives can be established faster to support clinical practices. Some Covid-19 questions we need answers for include, “Is Covid-19 causing a higher incidence of early pregnancy loss and preterm birth?”, “Is the prevalence of stillbirth higher in Covid-positive populations?”
Existing Covid-19 pregnancy registries rely on providers reporting patient information or patients self-reporting; both approaches struggle to successfully aggregate data at scale and with sufficient and diverse representation to limit or avoid bias. Registries tend to encounter enrollment challenges because physicians and practice staff are commonly overwhelmed and focused on delivering clinical care and not on registry building activity that will support future clinical research. Self-reported registries lack sufficient diversification for analysis because frequently patients of color or those facing barriers related to social determinants of health are distrustful of the healthcare system; the lack of trust often causes individuals to refrain from sharing their data. Finally, the current registries depend on a diagnosis of test-positive COVID-19 or suspicious symptoms (persons under investigation). How will we be able to determine the impact of asymptomatic infections on pregnancy outcomes without a comprehensive data set from which to analyze trends and patient experiences over time?
Health plans can aggregate some of the patient health data needed for real-world studies. However, despite many health plans having hundreds of thousands, if not millions of patients’ data on file, the data is often administrative, critical to paying claims, and not clinical. In situations where more advanced health plans receive patient clinical data from many types of providers, including Ob-Gyns, this data can come in many formats, making it challenging to compare and analyze across data sets. Imagine a health plan with two large Ob-Gyn groups in its network, sharing clinical patient data, but from two different EHRs (with the most valuable data in notes). Information is discordant, and data dictionaries are varied so that normalizing the two data sets into a single analyzable platform is one of the most significant challenges researchers looking for real-world evidence face.
There are companies in the market today solving portions of the real-world-evidence data set problem, so the good news is that comprehensive solutions, including ones specific to pregnancy, may be on the horizon. Companies like Covance, IQVIA, OptumHealth, Parexel, and HealthCore have droves of patient data, however much of it is administrative and lacks critical clinical detail for sound analyses. There are analytical platforms like Aetion, HealthVerity, and Definitive Healthcare that securely connect data sets and perform the analysis to answer clinical questions, but these companies often do not generate or have immediate access to the clinical data that they analyze. Then there are a few niche companies that rely on the EHR. A company like Elligo works to pull the data available within EHRs, while companies like Flatiron (Oncology) and Dorsata (Women’s Health) specialize in capturing clinical information at the point of care to power RWE-based studies.
Without the ability to aggregate representative, structured data in real-time for analysis, novel situations like the COVID-19 pandemic leave organizations and professional associations struggling to develop evidence-based guidance rapidly. Clinicians are left to determine “best practice” on their own, each provider having to use their best judgment uninformed by data. We all want to practice evidence-based medicine to deliver high-quality care for our patients. Without data, each provider makes decisions independently in a vacuum, causing stress and anxiety both for health care professionals and for patients. In 2020 there has to be a better way. If the COVID-19 pandemic can drive us to create a solution, the disruption may have at least one positive outcome.
Photo: Blue Planet Studio, Getty Images