Unlock Real World Evidence with Pragmatic Data Governance and Quality Practices

August 4, 2020
Kaitlyn Ramirez

How companies are leveraging Real Word Evidence to evaluate therapies under real world conditions in a broader population at a lower cost than using Randomized Clinical Trials.

Pharmaceutical companies capture and leverage massive, diverse data sets to power their Real World Evidence programs.Real World Evidence (RWE), as defined by the U.S. Food and Drug Administration (FDA), is the clinical evidence regarding the usage and potential benefits or risks of a medical product.1 In contrast with Randomized Clinical Trials (RCTs), RWE relies upon data captured in an observational manner in the patient’s natural, uncontrolled setting. Companies are augmenting their clinical trial information with RWE’s observational study of the patient journey and information about the day-to-day performance of drugs to support regulatory and policy decisions. Using these sources of data, we can evaluate therapies under real world conditions in a broader population and at a much lower cost than is possible with typical RCT. Companies are leveraging RWE as part of their strategic adaptation to the industry’s shift to specialty medications with narrow indications and smaller patient populations.

Regulators have stated their RWE position

Regulators are paying attention to pharmaceutical companies’ use of RWE more than ever before. The FDA stated their position in December 2018 using a strategic framework on RWE and a detailed commentary by agency leaders.2 In 2019, the FDA expanded the guidance and created a unified format for submitting new drug and biologic applications involving RWE. Pharmaceutical companies can expect regulators to increase examination and scrutiny of RWE included in regulatory submissions as part of their diligence to ensure accuracy, integrity and safety of data on patients who are not part of an RCT.

A prime example is a framework for evaluating companies’ use of data in producing RWE. The elements of the FDA’s framework approximate what companies need to do to ensure their data is of robust quality and is compliant with the regulators’ position. As part of their framework, the FDA stated their focus is exploring the potential of RWE to support regulatory decisions about drug product effectiveness. Specifically, the FDA has acknowledged that RWE has potential to be used to support changes to labeling drug product effectiveness, including adding or modifying an indication, such as a change in dose, dose regimen, or route of administration.3

The FDA states that companies must carefully assess both the reliability and relevance of RWE’s underlying data.4 This includes the methodologies that companies use to analyze the data to generate evidence to support products’ efficacy and safety. Application of data governance and data quality practices are key for achieving and maintaining compliance. Since RWE requires many data sources and data elements to build a patient’s longitudinal history in real world contexts, companies must apply these practices consistently, and with the capability to audit, along the data’s life cycle, from initial sourcing to its use in analysis workflows.

Safety and efficacy evidence require weaving complex patient data

Production of RWE requires the upstream integration of diverse data sets that weave together a patient’s events from different time periods and from various collection methods and device telemetries. Companies must understand the RWE’s underlying data because a data set’s type and attributes might be appropriate for only certain types of evidences and decision making. The data must represent the population of interest and needs to have enough data to power the study design after the inclusion/exclusion criteria has been applied, and that sufficient time has passed to measure the outcome of interest.

The non-experimental setting of electronic medical records (EHRs) represents a wealth of information about the patient’s real life context. Companies must have the capabilities to make sense of the complexities in this data type. For example, EHRs may be more appropriate for studying disease natural history and answering questions about treatment outcomes. EHR data containing a patient’s physician visit data and pharmacy dispensing data may be appropriate for studying a patient’s medication adherence, though there could be unexplained gaps in a patient’s history. An additional challenge is the lack of harmonized formats in the data that originate from the real world aspects of when the patient’s data was collected. One such example is when a drug name sometimes refers to the administered medication, and on other occasions it refers to the active chemical ingredient.

Additional data types outside of EHRs can provide information about how a drug has behaved in real-world conditions where population diversity, comorbidities and how a patient’s adherence variability come into play.Personal digital health apps and patient-generated data may be appropriate for understanding patient-reported outcomes.Administrative claims data may be suitable for pharmacoepidemiology studies since claims capture medically attended events helpful in defining observation periods.Patient registries may contain dimensions of disease registries and product-specific registries that add information about populations using specific products and assess long-term safety and adverse effects.

Pragmatic data practices can unlock RWE compliance

The complexities of RWE’s underlying data illustrate some fundamental ways it can block companies from achieving both critical regulatory compliance and sustainable business success. Data with unknown or poor quality and analysis methodologies lacking auditability not only prevent trust in RWE submissions content but also put data analysis processes at risk to deliver on time and to use expensive resources inefficiently. Companies can conquer these obstacles around the data trust and comply by employing pragmatic data governance and data quality methods.

A best practice for business stakeholders is to lead data governance and data quality efforts by making sure the regulator’s evidentiary needs and the RWE’s research design are in synch. When accomplished the business achieves alignment among cross-functional teams. This will create a clear, traceable lineage from RWE’s data sourcing to study execution to the evidence findings actionable in facilitating the decision-making related to the product. In execution, this means that business stakeholders need to lead discussions and agreements upon criteria for acceptance and rejection of data quality at all stages of data sourcing and evidence gathering. Business stakeholders also are the key translators to IT functions in explaining the technology requirements for ensuring accuracy and integrity of the data supplied to RWE gathering. Companies that succeed with these business responsibilities leverage programmatic data quality checks and efficient discoverability of data quality information. Companies with leading RWE programs have made use of current advances with data analytics and machine learning to enable automation in programmatic data quality and governance.

Another key stakeholder group involved in a company’s RWE program are the data scientists who deeply know the data and the links between data sets used in the evidence gathering. While the business stakeholders orchestrate sourced data according to regulatory compliance and the study design, the data scientists focus on understanding the complexities of the data and the options for constructing the patient’s journey, including how to evaluate the impact of gaps in the journey. Complexity can rapidly multiply because complete data could only be available for a subset of patients who may differ from other patients. Additionally, data scientists operate under constraint as the collection, storage and dissemination of personal real world health data is strongly regulated (e.g. GDPR). Industry experts have posited that it is unethical to create linkages across clinical datasets that do not already have intact linkage in place.5 Data scientists can be pragmatically equipped with high-quality, validated data dictionaries that remove the burden of constantly researching what the data is, where the data is coming from, and whether the data has the required quality to be fit-for-use for the needed evidence. With sufficient data dictionaries, data scientists can properly invest their skills and time into making sense of the complex patient data with statistical models and performing targeting analyses on the ever-expanding healthcare datasets.

In summary, pharmaceutical companies have seized the opportunity of RWE regulatory submissions and they continue to expand their usage of RWE. When companies employ RWE in their drug development or commercialization processes, they must comply with the regulators’ position. The FDA’s position on RWE indicates a shift to how data will be managed in a regulated environment. Companies should take action to empower their stakeholders involved in RWE to use pragmatic approaches in assessing their data environment rapidly, detecting and remedying data quality issues, governing data to institute trust, and applying improvements quickly.

References

  1. “Framework for FDA’s Real-World Evidence Program.” U.S. Food and Drug Administration, Dec. 2018, https://www.fda.gov/media/120060/download
  2. Ibid.
  3. Ibid.
  4. “Key Takeaways from FDA’s Framework for Real-World Evidence for Pharmaceuticals.” Covington Digital Health, 20 Feb. 2019, https://www.covingtondigitalhealth.com/2018/12/key-takeaways-from-fdas-framework-for-real-world-evidence-for-pharmaceuticals/
  5. “EHRs + Machine Learning Decipher Drug Effects In Pregnant Persons.” http://www.bio-itworld.com/2020/06/22/ehrs-+-machine-learning-decipher-drug-effects-in-pregnant-persons.aspx

Kaitlyn Ramirez, Experienced Manager, Digital Transformation & Management at Grant Thornton LLP