By Matthew Roe, M.D., Duke Clinical Research Institute
Real-world evidence is a hot topic in clinical research right now. But too often, the massive amounts of data now available from electronic medical records used in routine medical practice are being considered to be clinical evidence to support medical decision making. However, the process of transforming these large data sources into actionable evidence that can change clinical practice is a complicated but important endeavor.
Real-world evidence has been defined as healthcare information derived from multiple sources obtained from typical clinical research settings, including electronic health records (EHRs), claims and billing data, product and disease registries, and data gathered through personal digital devices and healthcare applications. 1
We are told we need this evidence, that it will help improve the clinical trial process by reflecting actual clinical practice and will generate better outcomes for patients. But to use this data, we must first understand how to obtain this information and which questions we can answer with it.
The 2009 Health Information Technology for Economic and Clinical Health Act mandated the adoption of electronic health records in the United States, giving healthcare providers and researchers enormous amounts of data to inform patient care and treatment decisions, profile patient outcomes, and facilitate pragmatic and comparative-effectiveness research studies. To bridge the gap from data to actionable evidence, the Patient-Centered Outcomes Research Institute (PCORI) invested more than $250 million in the development of PCORnet, the National Patient-Centered Clinical Research Network, which is a large national “network of networks” that collects data routinely gathered in a variety of healthcare settings, including hospitals, doctors’ offices, and community clinics, and transforms it into a common data model. By engaging a variety of stakeholders — patients, families, providers, and researchers — throughout the process, PCORnet empowers researchers to use data from these networks to answer practical questions that help patients, clinicians, and other stakeholders make informed healthcare decisions.
The ADAPTABLE Study: The Aspirin Dose Question
One PCORnet study testing this approach to transforming data into evidence is the ADAPTABLE (Aspirin Dosing: A Patient-centric Trial Assessing Benefits and Long-Term Effectiveness) trial, a pragmatic clinical trial funded by PCORI that will compare the effectiveness of two different daily doses (325 mg or 81 mg) of aspirin widely used to prevent heart attacks and strokes in individuals living with heart disease. ADAPTABLE is the first large-scale pragmatic trial conducted through PCORnet’s network of participating healthcare systems.
Despite aspirin being a key therapy for patients with cardiovascular disease to reduce the risk of a future cardiovascular event such as a heart attack or stroke, doctors variably recommend high-dose (325 mg) or low-dose (81 mg) aspirin with no clear preference for either dose. Furthermore, there is no consensus for a preferred dose of aspirin in the current American College of Cardiology (ACC)/American Heart Association (AHA) Guideline Recommendations for chronic cardiovascular disease, likely due to the relative lack of data from randomized trials regarding the optimal dose of aspirin for balancing the benefits versus the known risks of bleeding and gastrointestinal intolerance observed with aspirin.
Combining Traditional And Pragmatic Approaches To Identifying Potential Participants
Traditional clinical research trials are expensive, include only a narrow population of patients, are inefficient, and often don’t provide the critical evidence patients and clinicians seek to inform healthcare decisions. Recruitment is a challenge for all type of trials, traditional and pragmatic. Obstacles often include strict inclusion and exclusion criteria and burdensome study visits and procedures. To enhance study participation, start with the protocol design ─ minimize eligibility criteria and provide a streamlined schedule of assessment that reflects, as much as possible, routine clinical care.
In addition, avoid writing the protocol in a vacuum. Asking for feedback on the protocol design from both patients and clinicians is key to developing a protocol that reaches a broad population, can be operationalized easily, and ultimately provides results that inform practice.
ADAPTABLE employs both novel traditional and electronic approaches to identifying, recruiting, randomizing, and following participants, and it is disrupting the paradigms that have governed clinical trial conduct for many decades. Eight PCORnet clinical data research networks (CDRNs) representing more than 30 healthcare systems, one health plan research network, and one patient-powered research network are participating in ADAPTABLE. The CDRNs use programming specifications that define patient inclusion/exclusion criteria for the trial (termed a computable phenotype) to query their EHR data to facilitate widespread screening of potentially eligible patients. Once eligible patients are identified, a variety of outreach methods are employed, including direct mail, email, health system portal messages, best practice alerts to providers, and tablet-based recruitment during clinic encounters.
Conducting Pragmatic Clinical Trials In Learning Health Care Systems
Once potential participants are identified and contacted, they are provided a special code and directed to the ADAPTABLE web portal for confirmatory screening, electronic consent, and self-randomization. Enrolled participants return to the portal every three to six months to complete study surveys that ask about hospitalizations for myocardial infarction (MI), stroke, or major bleeding.
In addition to patient-reported outcomes, networks perform routine queries of the PCORnet Common Data Model (CDM) to capture and classify endpoints for hospitalizations for MI or stroke as well as death events. The PCORnet CDM is based on the FDA’s Sentinel Initiative CDM and has been informed by other distributed data initiatives. Each PCORnet partner network maps data to the same consistent format, creating a platform that enables much more rapid responses to research-related questions.
Data from hospitalizations not captured in the PCORnet CDM comes from surveillance of data from the Centers for Medicare and Medicaid Services (CMS), from private health plan networks, and from patient reported outcomes that are confirmed by contact with personnel at a centralized call center. Ascertainment of death is derived from querying the PCORnet CDM and Social Security Administration (Medicare beneficiaries), and from information gathered from the study call center’s contacts with participants who miss their electronic follow-up visits with the trial web portal.
A Signature Study Of Collaboration
In addition to pragmatic operational approaches, ADAPTABLE provides a model of engagement with all stakeholders — patients, clinicians, researchers, and professional associations. Feedback from a patient representative group, the ADAPTORS, has been crucial to the success of ADAPTABLE. The ADAPTORS are patients with heart disease who are members of the study team who shape the direction of the study through their participation in committee meetings, working group calls, and local and national awareness events, and by reviewing the study portal and materials.
Although advances in technology allow patients to self-consent and self-randomize with the ADAPTABLE web portal, patients still want to hear about and discuss participation in clinical trials with their clinicians. A significant portion of study outreach is directed at educating clinicians about the evidence gaps and uncertainty for delineating the optimal dose of aspirin for cardiovascular disease. Once clinicians hear about ADAPTABLE, they typically have been very excited to learn more about the study and want to collaborate and contribute to patient enrollment.
Both the American Heart Association and American College of Cardiology are partnering with ADAPTABLE to help spread awareness of the study. ADAPTABLE will provide results to help clarify practice guideline recommendations authored by these associations as well as an infrastructure for clinician-scientists to ask important clinical questions in a cost-effective, patient-centric manner.
Lessons Learned So Far
The experiences with the trial thus far have been very positive and have reinforced the need for simplicity in clinical trial conduct. Lessons learned to date include the following.
First, make the study design as simple as possible and avoid the temptation to introduce more complexity. If you cannot showcase the inclusion/exclusion criteria on one slide, the study is probably too complicated.
Second, make sure the research question is not only important to clinicians but also that the trial conduct and execution can be implemented seamlessly into clinical practice without disrupting patient flow or clinical care.
One example of a trial being implemented into clinical practice is ADAPTABLE’s use of best practice advisories (BPAs). Originally intended to assist clinicians in improving performance and patient outcomes for certain high-priority health conditions, this decision support tool can also be used to identify potential research participants. The benefit of BPAs for research is that clinicians are notified in real time during clinic encounters if their patient is eligible for ADAPTABLE. They are provided a brief summary of the study and can choose to discuss the study with their patient or place a referral for the study team to contact the patient. What we have learned from networks in ADAPTABLE who have worked with their EHR team to program BPAs is that BPAs work well when the alerts are programmed as a pop-up note that appears on the clinician’s screen versus being posted on a supplemental screen that a clinician would need to navigate to read the alert in a list of other alerts. In addition to visibility, “BPA fatigue” is another noted disadvantage, the result of providers being bombarded with numerous alerts that have been deployed presumably to improve clinical care. Providers typically become desensitized and may overlook the research BPAs in this context.
What other questions can be answered with the ADAPTABLE model? Could other questions about diet, exercise or medication adherence relevant to many chronic diseases and conditions beyond cardiovascular disease be asked using this type of research model as well? Aspirin is not the only example in which clinicians often have to make anecdotal treatment recommendations based on their personal preference or what they think is best because of the lack of evidence. ADAPTABLE provides a new model to answer questions that are important to patients and their clinicians.
About The Author:
Matthew Roe is the faculty director of the Global Outcomes Commercial Megatrials group at the Duke Clinical Research Institute and professor of medicine in the Division of Cardiovascular Medicine. He received his medical degree and his Masters of Health Sciences in Clinical Research from Duke University School of Medicine.