Leveraging Real-World Data for COVID-19 Research: Challenges and Opportunities
by Matthew W. Reynolds, Jennifer B. Christian, Christina D. Mack, Marni Hall and Nancy A. Dreyer
Due to the rapid and severe onset of the COVID-19 pandemic, unprepared medical systems worldwide found themselves overwhelmed, with traditional public health approaches disrupted. Critical information – including modes of transmission, clinical presentation of disease, and clinical outcomes post infection – is still elusive. Characterizing the sudden onset of a highly prevalent disease is not straightforward as the condition is novel, and new findings are emanating from numerous unvalidated sources.
Data and information have at times been inaccurate, inconsistent, and/or heterogenous, due to limited testing capacity and yet-to-be standardized testing criteria. Additional variability is introduced by shifting prioritization of populations being tested, from a focus on the sickest patients to be more inclusive of the most high-risk populations; this is further confounded by changing guidelines on when individuals should seek medical attention. Questions continue to mount around the clinical presentation of the disease across geographic regions, treatment effectiveness, and clinical outcomes of COVID-19.
A systematic approach to generating information to support clinical and public health decision making is needed. Real-world data (RWD), captured directly from patients or through medical claims or electronic medical records (EMRs), has the potential to inform COVID-19 research in ways randomized clinical trials (RCT) will not (see Figure 1). For example, epidemiological tactics such as contact tracing have been employed to minimize further spread and save lives while effective treatments and vaccines are developed.
This paper aims to address core challenges facing the use of existing data for COVID-19 research, with emphasis on epidemiologic methods and data quality. We offer insights and potential solutions for leveraging and interpreting RWD about COVID-19.
Epidemiologic Methods Considerations
A major goal of human research on COVID-19 is to find out whether any currently marketed pharmacologic interventions could be effective in preventing infection, or in lessening the severity of the infection. This includes both progression to and treating severe acute respiratory syndrome-coronavirus-2 (SARSCoV- 2) which causes people to be hospitalized and often results in death. In fact, RCTs are now underway to evaluate existing and new therapies (e.g., RECOVERY trial UK,1 PCORnet HERO study2). While RCTs are pragmatic, there are opportunities for non-randomized RWE to provide information in the interim as well as to answer additional real-world questions that are not addressed in the studies. Such investigations not only need reliable data of good quality, but also need to be approached with strong methodological considerations.3
Important public health assessments that could be answered now with non-interventional approaches using RWD include:
- influence their severity.
- Other covariates and outcomes may also be discovered depending on the specific study objectives, including the unintended effects of social isolation, mental health issues, and domestic violence. Other factors to consider are the consequences of reduced access to medical Identifying the prevalence and incidence of cases
- Describing the medications prescribed and when are they given in the patient’s clinical course
- Characterizing the use of pharmaceutical and non-pharmaceutical products for intended for prophylaxis or curative treatments, recognizing the need to distinguish product use in relation to symptom onset since some drugs can be used for both purposes.
- Classifying and describing treatment use based on severity level.
- Examining treatment heterogeneity to better understand patient characteristics, that is, determining cases for which, or when, medicines are most effective, since not everyone will benefit to the same extent.
- Understanding concomitant medication use and potential interactions (e.g., which prescription medications, including biologics, that may confer some protective benefits as well as antihypertensives and antidiabetics, that may confer added risk; likewise, an understanding and assessing nonprescription medications, and others).
- Understanding the risk of serious clinical outcomes in patients and the factors that care for non-COVID-19 conditions, e.g., allowing follow-up cancer visits to be delayed or having longer intervals between treatments requiring infusions (e.g., some biologic therapies), or, the potential risks of non-adherence to treatments, such as interruptions in diabetes and/or hypertension treatments.
Early research has suggested benefits for some medications as treatment or prophylaxis for COVID-19,4 while others have noted increased risks of COVID-19 related to other treatments such as antihypertensives.5 With clinical practice patterns and patient medication-taking behavior likely to change in these times (e.g., reduced access to care, avoidance of seeking medical help, etc.), there is also strong potential for unintended consequences that could drive an increase in the incidence of illnesses related to untreated conditions such as diabetes (diabetic retinopathy) and hypertension (cardiac events), etc. We also anticipate seeing other indirect effects of COVID-19 such as an increase in domestic abuse and psychiatric conditions, including suicide, perhaps stimulated in part by social isolation or increased anxiety.
Targeting Patients with COVID-19 From an epidemiologic perspective, the first step in studying COVID-19 is to identify the patients of interest. Some think of COVID-19 as a disease in and of itself, regardless of symptoms and severity. While many people with a documented infection of SARS-Cov-2 do not exhibit symptoms,6 the primary interest – our problem to solve – is altering the clinical course to avert serious adverse respiratory distress syndrome from the SARS-Cov-2 virus itself, which requires hospitalization and can be fatal in many cases.7 Beyond recording and reviewing consequences of life-saving and life-altering treatments for COVID-19, we also have a critical interest in managing patients with symptoms serious enough to seek medical care. Another important component of managing patients is screening for and then understanding populations of people who are asymptomatic or those with subclinical manifestations but who are documented for COVID-19 exposure. The goal in these cases is to understand the characteristics associated with having less severe and/or no symptoms, and to protect others from exposure to this population. Many of the activities are confounded by the changes in how people are seeking medical care in light of this epidemic, particularly the large move to telemedicine which forces medical care providers to shift to evaluating and treating patients without a traditional, hands-on examination.
Identification via COVID-19 Laboratory Testing
In clinical practice, the identification of a COVID-19 patient is confirmed via a positive laboratory test. Three kinds of tests are currently available for COVID-19: nucleic acid-based viral tests that determine if a patient has a current infection; an antigen test to detect fragments of viral proteins from a patient’s nasal cavity using swabs; and antibody tests that indicate whether the patient has had a recent or previous infection. Unfortunately, these tests (including component parts, such as swabs) have not been readily available, and many health care practitioners have had to deal with either a low supply or no supply to support their growing pool of potential COVID-19 patients. To add even more complexity to the shortage of COVID-19 tests, there are a wide number of tests on offer. Today, more than 230 test developers have alerted the Food and Drug Administration that they are requesting emergency use authorization (EUA) for their COVID-19 diagnostic and/ or antibody tests, and twenty EUAs have been granted; additionally, at least 110 laboratories around the country are also using their own tests.
The rapid development and response to making tests available comes with significant heterogeneity in the accuracy of testing results within and across each of the commercial kits. A test could give a false negative, meaning that for, say, the serology test, a patient has the Covid-19 antibodies, but the test mistakenly reports a negative. A test can also give a false positive reading, meaning that the patient does not have antibodies but the test reports that the patient has antibodies. Given the recency of antibody tests made available, there is limited information available on the accuracy of testing, and even less information on how to interpret the findings. The FDA had loosened its rules for testing citing the COVID-19 as a public health emergency and relied on manufacturers to conduct their own validations. Given the concerns raised above, the FDA has now revised its policies to impose stricter review and validation processes.8-10
For RWD that includes these laboratory results for COVID-19 testing, researchers need to keep in mind that there will be many more people infected by COVID-19 who never receive a test – hence, there would be no results on record (positive or negative). Even with extensive testing, capturing and ascertaining tests results is complicated by recently available at-home kits and drive-thru facilities where results may not make it to a physicians’ EMR, let alone be formatted for use in RWD analyses of COVID-19.
Identification by Diagnosis Codes and Symptoms
In the absence of COVID-19 testing results, there are other approaches that identify patients who currently have or had the disease. Most RWD is derived through encounters with the medical care system. Patients of interest may be identified by searching an ICD-10 diagnosis code, or when those are not specific enough, potentially via some combination of clinical data elements like lab tests, procedures, medications, or clinical text that may provide descriptive specificity in identifying cases and distinguishing them from non-cases.
A new ICD-10 was created and released for use on April 1, 2020 (ICD-10 U07.1 COVID-19).Since the code did not exist during the earlier months of the pandemic, these results may take some time to be fully integrated into medical coding with medical claims data and EMR. Prior to creation of this code, providers were in some cases coding general conditions (e.g., “other viral pneumonia”, “unspecified acute lower respiratory infection”) possibly in tandem with B97.29 “Other coronavirus as the cause of disease classified elsewhere.”11,12
However, guidance stipulated that “suspected,” “possible,” or “probable” COVID-19 should not receive B97.29, and instead diagnostic coding assigned to explain the reason for the encounter; in the absence of available screening, early attributable cases will be missed. For prudent use of RWD from medical claims or EMR, researchers are warranted to acknowledge the early unavailability of this code, the workarounds utilized, and the gradual standard use and screening availability of the COVID-19 code.
In lieu of a specific code, it may be possible to identify patients via diagnosis codes of patient symptoms. The Centers for Disease Control and Prevention recently released the most common symptoms that may appear two to fourteen days after exposure to the virus, including cough (ICD-10 R05), shortness of breath (R06.02) or difficulty breathing (R06.00), or at least two of any of these symptoms: unspecified fever (R50.9), chills (R68.83), muscle pain (M79.1), headache (R51), sore throat (R07.0, J02.9), or a new loss of taste (R432) or smell (R43.0). Issues with trying to identify COVID-19 patients by their symptoms alone are described below:13
- Mild symptoms experienced by many patients are not typically drivers of billing claims; a more severe/serious diagnosis code may be used as the reason for visit and documented in the medical claim without mention of each of the individual symptoms
- Many COVID-19 symptoms may not necessitate medical care; a patient may feel fatigued or have a cough and treat themselves at Hence a full reliance on this approach will result in a likely underestimate of the full COVID-19 patient population, with a bias towards the more severe cases.
- These mild symptoms are not specific to COVID-19, widely are variable in presentation and combination, and, unfortunately, overlap with symptoms from very common conditions such as influenza, strep throat, and the common cold. Based on this symptomatology alone, it would be very difficult to differentiate COVID-19 patients from those patients with other potential illnesses.
Another approach to identify those who may be more likely to have been diagnosed as COVID-19 patients would be to expand the diagnosis codes from the COVID-19 U07.1 code to similar conditions which may have been coded for billing purposes (or prior to the full implementation of the new U07.1 code), similar to how COVID-19 patients would be ascertained prior to code creation.
Per guidance from the CDC,14 a broader definition could include adding codes such as B97.29 (“Other Coronavirus as the cause of disease classified elsewhere”) or B34.2 (“coronavirus infection, unspecified). This broader definition would allow for a number of coronavirus potential diagnosis codes (COVID-19 specific and non-specific). Further, one could include any of the following codes: pneumonia (J12.89), acute bronchitis (J20.8, J40), lower respiratory infection (J22, J98.8), and ARDS (J80) as potential inclusionary diagnosis codes, and further expanding to include any patients who have a billing code for a COVID-19 test ordered or a test taken (HCPCS U0001 or U0002 and CPT code 87635).
This approach broadens the inclusion criteria to more specific diagnosis codes while incorporating a variety of potential codes to represent the COVID-19 diagnosis. Depending on the subset of patients of interest, additional restrictions and/or definitions could be applied such as hospitalized COVID-19 patients (e.g., looking for these diagnosis/procedure codes in an inpatient setting).
It is also possible to leverage medical encounter data along with richer clinical textual information from within the EMR to identify and/or confirm true COVID-19 patients more accurately. This type of EMR data within Integrated Health Systems has been reported across various venues such as Johns Hopkins University,15 while other organizations use the COVID-19 Evidence Accelerator supported by the Reagan-Udall Foundation for the Food and Drug Administration and Friends of Cancer Research.16
These data provide us insights into the COVID-19 burden of disease as well as treatment patterns and clinical outcomes from within those clinical systems. While these insights are valuable, and often clinically rich, the data sets are typically drawn from smaller patient sample populations (e.g., localized to regions or health systems); hence, the samples and data may not represent the general patient populations, clinical treatment patterns, and/or risk of outcomes outside of that health system.
Examining Health Outcomes – Ongoing Surveillance
Once people infected with COVID-19 are identified and treated, a common and clinically important question remains – what happened to them? Were they admitted to a hospital or quarantined at home? Did they recover successfully or have poor clinical outcomes? Before ascertaining these outcomes, we also need to assess the validity of the data source before we measure and report results.
It is important to examine the RWD source and determine whether the source contains the full continuum of care (outpatient tracking, prescriptions, inpatient care, emergency room visits, etc.). For example, datasets that identify COVID-19 patients via diagnosis codes and laboratory results may not have complete data for all their medical care (e.g., ambulatory care or billing data may not include information about whether the patient was hospitalized recently). Further, if the data source includes most, if not all of the continuum, how confident can one be that the source has the complete clinical data records for the patients of interest? It is essential to be assured that the data source of interest is “fit for purpose” for the selected research objective.
Many clinical outcomes have been consistently noted in COVID-19 research and should be examined, including:
- Hospitalization and length of stay
- Disease progression as measured by
- Ventilators; days on ventilator
- Movement within the hospital (e.g., ER to admission to ICU, )
- Blood clots (e.g., stroke, heart attacks, pulmonary embolism, deep vein thrombosis)
- Kidney failure/dialysis
For hard clinical outcomes that are easy to measure objectively, such as a hospitalization, the key focus is ensuring that this information exists, it is accessible/identifiable, and it is complete in the dataset of interest. We can access many large, real-world medical claims from databases that cover 100+ million patients, and while they may be strong and deep for one type of medical care such as prescriptions, they may not capture other healthcare services completely. Similarly, very rich EMR datasets can be used to track a patient through an integrated health system (from outpatient to inpatient services), but patients may seek care outside of the health system, care that would not be visible in the EMR and hence would be “unknown/missing” to researchers’ eyes.
Such data are important to know at the outset if a key outcome variable likehospitalization is significantly incomplete in the dataset. But if so, can you consider other data source options or determine the best approach on how to compensate or incorporate that limitation into your study conclusions? When working with a medical claims dataset, it would be more reliable to leverage a “closed” system that has complete claims for a single patient via their specific insurance plan and/or employer. Unfortunately, the closed plans typically require 90 days or more to curate; hence, the general medical claims data may be more accessible on a real-time basis but less complete. Regardless, these data still may provide insight into what is happening with the COVID-19 patients.
While incomplete hospitalizations in a data source may present serious issues in calculating risk or rate of hospitalization(s) as an outcome, it is possible to derive insights from the hospitalizations that are measured to characterize: the length of stay within the hospital; the care provided (e.g., procedures/ treatments that indicate poor outcomes); and evaluate pre-hospitalization factors that may influence the hospitalization results. While in the hospital, it may be possible to identify more severe cases/outcomes by the presence of codes/notes of ventilator usage (depending on the level of detail captured). The length of stay may be indicative of the severity of the case and/or poorer outcomes, and some data sources will allow you to track a patient from bed to bed and unit to unit within the hospital (or across hospitals). The location of a patient within the hospital, such as within an Intensive Care Unit (ICU), may be used as a potential differentiator for severe vs non-severe cases. Of course, none of these resources address the functional status of people who survive hospitalization, a topic also of wide interest.
Mortality is a metric often reported at the national and state levels currently for assessment of COVID-19 burden. In RWD, however, mortality is not as simple a metric to ascertain as one might assume since most RWD sources are not linkable to national mortality data. In the US, the Social Security Administration data is likely the best and biggest option, but it includes a time lag, and the completeness of the mortality reporting has been decreasing over recent years.18,19 Some researchers are supplementing sources with other obituary data, but there is no single option and/or best method currently for which a researcher can do this easily. The National Death Index is used in many research studies, but this process requires submitting a patient list with patient identifying information (for a fee) to receive vital status for all people on the submitted list. Inherent lag time for any of these options impedes assessments for a broad multitude of RWD sets, hence making individual patient analytics almost impossible for sufficient real-time tracking mortality in COVID-19 patients.
Integrated health systems and some hospital data sources may have the ability to identify patients who die within the hospital. Although not typically 100% complete and reliable, it would be expected that the data for these patients may be more complete and reliable, especially given the focus on COVID-19 nationally. Hospital data sources may still miss information on deaths outside of the hospital but at least provide a viable metric for “in- hospital mortality.”
Considerations about healthcare access and utilization in the COVID-19 context
One unusual feature of this epidemic is that people may now be reluctant to visit hospitals and other medical care facilities, likely for fear of increasing their chances of exposure to COVID-19. Due to over-crowded healthcare systems, patients with mild cases may not be encouraged to come in for follow-up when they are doing well. This phenomenon may prevent researchers from being able to identify clinical outcomes of their patients if they never make it into the medical system. Similarly, the less-severe cases that resolve untreated while a patient remains at home would never be counted. Assessment of the full picture of cases, symptoms, and outcomes would likely require a RWD set with a broader reach that includes proactive and focused data collection from all patients affected by COVID-19. Many efforts are underway to collect this information, including the CARE Project (www.helpstopCOVID19.com) or collecting information on specific groups of patients such as those with cancer19 (ASCO registry) or celiac disease (SECURE-Celiac).20
Other outcomes that may be important to assess, but unlikely to be included in most real-world or clinical data sources may include things like time and ability to return to work, quality of life assessments including post-hospitalization, length of time to symptom resolution, economic burden, mental health impact, and others that would require proactive data collection directly from patients/caregivers.
Ultimately, the best approach and data source depend on the questions of interest – incidence, prevalence, characterizing symptoms, treatments – but resolving these questions may involve leveraging a combination of approaches. Considering the rapid accumulation and changing evidence that is being developed and assessed, the definitions and approaches to RWD should be constantly reviewed, scrutinized, and modified to accommodate all the new insights being generated. While it is critical to characterize the sources of data and understand the limitations, these non- interventional studies using RWD should be viewed as complementary to traditional trials and other studies.21 With good epidemiology methods, insights can be derived efficiently and reflect the clinical care received on a broad population. Conclusions need to be placed in the proper context of the data, clinical treatment patterns within the framework of each RWD data source (e.g., case identification, treatment patterns, and outcome rates in New York City may differ notably from those in rural Texas).
Moreover, it may be time to reconsider the exclusive reliance on EMRs and administrative health insurance claims. As digital tools become more prevalent, actively collecting data from clinicians and/or patients should become easier. In the context of a pandemic like COVID-19, however, it is unwise to overburden the health system with additional requests for health care providers. In contrast, information is being contributed by people in communities in response to invitations delivered through social media, patient advocacy or affinity groups, or other membership organizations. Considering the high level of public health interest in COVID-19, people appear eager to track and retain testing information as well as to describe their symptoms and concomitant medications. Further, in some settings, it is possible to link patient-reported information with their corresponding RWD from health plans, clinical records, etc.22 This direct-to-patient approach may yield engaged participants who will be likely to provide additional information about their experience alongside other RWD sources to answer a broad range of clinical questions.
- A wide range of RWD options can be used to evaluate the epidemiology, treatment, and clinical outcomes associated with COVID-19. The methods to examine COVID-19 in these data sources are rapidly being developed and implemented, and with good scientific assessment and methodology, these data can be employed eﬀectively for important public health
- Each RWD source will come with unique issues around definitions of disease, treatment, and A researcher must vet each data source thoroughly to determine if it is fit for purpose for the research question(s) at hand.
- RWD should be leveraged to provide rapid assessment of real-world clinical evidence that can be used to better identify high risk populations, eﬀective prevention/treatment strategies, and to inform formal clinical.
Matthew W. Reynolds, PhD, FISPE
Vice President, Real World Evidence, IQVIA
As a member of the Center for Advanced Evidence Generation in Real-World Solutions, he designs innovative solutions for real-world evidence on effectiveness and safety, reporting to the Chief Scientific Officer. He concentrates on studies that are enriched by combining primary and secondary data, including pragmatic trials. Dr. Reynolds brings more than 20 years of diverse experience in non-interventional research, including serving as Vice President of Epidemiology at Evidera, where he led a large team conducting real world data projects and literature reviews, and managed several large client initiatives and partnerships. He has a deep expertise in the usage of real-world data to address questions of eff ectiveness, safety, and value of pharmaceutical products across a wide range of therapeutic areas. He has held positions with AstraZeneca in cardiovascular epidemiology and Pharmacia as a pharmacoepidemiologist and senior analyst. He also has completed numerous assignments for the School of Medicine at the University of Maryland (Baltimore). Dr. Reynolds has broad therapeutic experience and methodologic experience and has published more than 50 peer-reviewed papers. He founded and has led the successful Database Special Interest Group for the International Society of Pharmacoepidemiology (ISPE) where he served for three years as an elected executive leader in the role of Vice President of Finance. Dr. Reynolds earned his graduate and doctoral degrees in epidemiology and preventive medicine from the University of Maryland at Baltimore in January 2000.
Nancy Dreyer, PhD, FISPE
Chief Scientific Oﬃcer and Senior Vice President, IQVIA
Nancy Dreyer is chief scientific oﬃcer and SVP at IQVIA, where she works on generating real-world evidence for regulators, clinicians, patients and payers through pragmatic randomized trials and non-interventional studies. She is a widely known thought leader, most recently for regulatory use of real-world evidence for label expansions. Her accomplishments include creating the GRACE Checklist, the only validated checklist for measuring the quality of observational studies of comparative eﬀectiveness, and serving as a senior author of 4 editions of “Registries for Evaluating Patient Outcomes: A Users Guide” published by the US Agency for Healthcare Research and Quality. She has authored more than 100 publications in peer-reviewed journals on diverse topics and is frequently quoted in the media. She also holds a position as an Adjunct Professor of Epidemiology at the University of North Carolina School of Global Public Health. She is a Fellow of both the International Society of Pharmacoepidemiology and DIA and is a member of DIA’s Scientific Advisory Committee. In 2018 was appointed to the US Patient-Centered Outcomes Research Institute Clinical Trials Methods Advisory Panel and has been a standing consultant to the US National Football League Health & Safety Executive Committee since 2013. In 2019, she was designated a key IQVIA personnel for the FDA Sentinel Community Building Outreach Center which was awarded to IQVIA in collaboration Deloitte. Twice named to PharmaVOICE’s list of 100 most influential and inspiring individuals in life sciences, in 2019 she received DIA’s Global Inspire Award for Author of the Year for “Advancing a framework for regulatory use of real world evidence: When real is reliable,” the most downloaded publication in 2018 in Therapeutic Innovation & Regulatory Science.
Jennifer B. Christian, PharmD, MPH, PhD, FISPE
Vice President of Clinical Evidence, IQVIA Real-World Solutions Jennifer Christian is Vice President of Clinical Evidence & Epidemiology at IQVIA, an adjunct faculty member at Weill Cornell Medical College, and an Anniversary Fellow of the Institute of Medicine. At IQVIA, her research focuses on strengthening clinical eﬀectiveness and safety evaluations of treatments and advancing the use of RWE for regulatory decision- making. She has led the design and conduct of registries, direct-to-patient extension studies, pragmatic trials, and external comparator studies for new drug applications and label extensions. She is also actively engaged on projects through Friends of Cancer Research, Duke Margolis Health Policy Center, and the National Academy of Medicine. She is a graduate of the UNC- Chapel Hill School of Pharmacy, UNC School of Public Health, and Brown University School of Public
Christina Mack, PhD, MSPH, Vice President, Epidemiology and Clinical Evidence, IQVIA
Adjunct Faculty, UNC-Chapel Hill
Dr. Mack is Vice President of Epidemiology and Clinical Evidence at IQVIA where she oversees development of large evidence platforms and novel studies that augment primary data collection with existing data. Dr. Mack, a recognized expert in eﬀectiveness studies, orthopedic injury research, and epidemiologic methods, is formally trained in public health and computer engineering. She holds Ph.D. and master’s degrees in Epidemiology from the University of North Carolina at Chapel Hill and a Computer Science Engineering degree from the University of Notre Dame. Her interests include the curation and use of electronic health records and claims data, with focus on complex study designs, machine learning, and methods such as external comparators, propensity scores, and data linkage and enrichment. Dr. Mack brings work experience from Johnson & Johnson, GlaxoSmithKline, IBM and non- profit global health and governmental organizations. She is a frequently invited speaker at industry and academic forums on epidemiologic methods and novel study designs, as well as careers in research, has published in leading journals and authored 4 chapters in the Agency for Healthcare Research and Quality (AHRQ) landmark publication “Registries for Evaluating Patient Outcomes: A User’s Guide” on the topics of designing registries for studies of medical devices, selection of data elements for observational research, missing data, and research networks. Dr. Mack serves as Co-Chair of the MDEpiNet Scientific Oversight Committee, Co-Chair of RAPID Safety Signal Discernment and Biostatistics workgroup, Chair of the International Society for Pharmacoepidemiology Medical Devices Special Interest Group, is an Advisory Board member for the Carolina Health Informatics Program and holds an academic appointment as adjunct Assistant Professor of Epidemiology at the University of North Carolina at Chapel Hill.
Marni Hall, PhD, MPH, Vice President of Clinical Evidence, IQVIA
Marni Hall leads IQVIA’s US Regulatory Science and Strategy Consulting business, helping clients integrate new data, methods, and tools into their development programs, through education, strategic planning, and implementation. Prior to launching this new function, Dr Hall was Vice President of Clinical Evidence in the Center of Advanced Evidence Generation at IQVIA. In this role she leveraged her expertise in regulatory science, drug safety, and patient-centricity, providing scientific oversight and strategic direction on the expanded use of real-world evidence for regulatory and other uses. She led internal and external teams in regulatory science activities such as data validation activities and the development of novel study designs, as well as collaborates with R & D and technology services groups to integrate solutions across the company. Dr Hall works closely with key scientific thought leaders on advancing the regulatory science of novel trial designs, and the expanded use of real-world data.
A research scientist by training, Dr. Hall spent nearly two decades at the intersection of science and policy. After serving as Program Director in the Public Health Group of External Medical Aﬀairs at Pfizer, Hall joined the FDA’s Oﬃce of Planning and Informatics in 2008 as a Principal Analyst. In this role, she initiated and led the development of the data standards plan for the Center for Drug Evaluation and Research (CDER). In 2011, she was appointed Director of Regulatory Science within the Oﬃce of Surveillance and Epidemiology (OSE) for the CDER, where she worked through 2016, leading safety science research priorities and activities at the Center- and Agency-level, developed new Agency and external collaborations including identification, evaluation and implementation of new data sources for use in regulatory decision making, and oversaw multiple contracts to conduct hypothesis-testing safety studies. Hall also became an expert at sourcing and analyzing big data sets, including adverse event reports, claims, and other data useful to risk assessment and risk management activities. Then, at PatientsLikeMe, Dr. Hall served as the Senior Vice President of Research and Development, Informatics, and Policy, where she oversaw delivery of commercial research programs and execution of an agenda for use of patient-generated health data in clinical and regulatory decision making.
Dr Hall was an adjunct professor at Columbia University for 10 years and is currently a Trustee of Worcester Polytechnic Institute, serving a Chair of the Academic Planning Committee on Distinctive Graduate Education, Vice Chair of the Academic Planning Committee, and member of the Audit, and Economic Development committees.
Dr Hall holds Bachelor of Science degrees in chemistry and in society, technology, and policy from Worcester Polytechnic Institute. She also holds a master’s degree in public health and a master’s degree in biochemistry and a PhD in toxicology from Columbia University.
- RECOVERY Randomised Evaluation of COVID-19 Recoverytrial.net
- HERO Healthcare Worker Exposure Response & Outcomes https://heroesresearch.org
- Understanding the Need for Non-Interventional Studies Using Secondary Data to Generate Real-World Evidence for Regulatory Decision Making, and Demonstrating Their Credibility (2019, November 25); https://healthpolicy.duke.edu/publications/understanding-need-non-interventional-studies-using-secondary-data-generate-real-world
- Wu R, Wang L, Kuo HD, Shannar A, Peter R, Chou PJ, Li S, Hudlikar R, Liu X, Liu Z, Poiani GJ, Amorosa L, Brunetti L, Kong An Update on Current Therapeutic Drugs Treating COVID-19. Curr Pharmacol Rep. 2020 May 11:1-15.
- Mali SN, Thorat BR, Chopade A Viewpoint on Angiotensin- Converting Enzyme 2, Anti-Hypertensives and Coronavirus Disease 2019 (COVID-19). Infect Disord Drug Targets. 2020 May 10.
- Bendavid E, Mulaney B, Sood N et COVID-19 antibody seroprevalence in Santa Clara County, https://www.medrxiv.org/content/10.1101/2020.04.14.20062463v2
- Lipsitch M, Swerdlow DL, Finelli Defining the epidemiology of Covid-19- studies N Engl J Med 2020; 382:13: 1194-1196
- There are many COVID-19 tests in the US – how are they being regulated? https://theconversation.com/there-are-many-covid-19-tests-in-the-us-how-are-they-being-regulated-134783
- Policy for Coronavirus Disease-2019 Tests During the Public Health Emergency (Revised). https://wfda.gov/regulatory-information/search-fda-guidance-documents/policy-coronavirus-disease-2019-tests-during-public-health-emergency-revised
- Insight into FDA’s Revised Policy on Antibody Tests: Prioritizing Access and https://wfda.gov/news-events/fda-voices/insight-fdas-revised-policy-antibody-tests-prioritizing-access-and-accuracy
- ICD-10-CM Oﬃcial Coding Guidelines – Supplement Coding encounters related to COVID-19 Coronavirus Outbreak Eﬀective: February 20,https://www.cdc.gov/nchs/data/icd/ICD-10-CM-Oﬃcial-Coding-Gudance-Interim-Advice-coronavirus-feb-20-2020.pdf
- Billing for COVID-19 Evaluation and https://whca.wa.gov/assets/billers-and-providers/billing-for-COVID-CPC-03-2020.pdf
- Symptoms of https://www.cdc.gov/coronavirus/2019-ncov/symptoms-testing/symptoms.html
- News: CDC releases coronavirus ICD-10 supplement coding guidelines, March 5, 2020 – CDI Strategies – Volume 14, Issue https://acdis.org/articles/news-cdc-releases-coronavirus-icd-10-supplement-coding-guidelines
- COVID-19 Dashboard by the Center for Systems Science and Engineering (CSSE) at John https://coronavirus.jhu.edu/map.html
- COVID-19 Evidence https://www.focr.org/covid19
- Reynolds MW, Bourke A, Dreyer Considerations when evaluating real-world data quality in the context of fitness for purpose. Pharmacoepidemiol Drug Safe 2020;1-3. DOI: 1002/ pds.5010
- Navar AM, Peterson ED, Steen DL, Wojdyla DM, Sanchez RJ, Khan I, Song X, Gold ME, Pencina Evaluation of Mortality Data from the Social Security Administration Death Master File for Clinical Research. JAMA 2019 Apri 1:4(4):375-379
- ASCO Survey on COVID-19 in Oncology (ASCO) https://www.asco.org/asco-coronavirus-information/coronavirus-registry
- Now Open: Registry for People with Celiac Disease who are Diagnosed with COVID-19. https://celiac.org/about-the-foundation/featured-news/2020/03/now-open-registry-for-people-with-celiac-disease-who-are-diagnosed-with-covid-19/
- Dreyer, , Hall, M. & Christian, J.B. Modernizing Regulatory Evidence with Trials and Real-World Studies. Ther Innov Regul Sci (2020). https://doi.org/10.1007/s43441-020-00131-5
- IQVIA COVID Active Research Experience (CARE) helpstopCOVID19.com; listed on https://clinicaltrials.gov/NCT04368065