Real-World Data: What Is It and
Why Does It Matter?

Chapter 2

Real-World Data Ecosystem

Numerous types of patient data gathered from multiple sources can be useful for analyzing RWE. To develop a deeper understanding of how RWD can be used and its value in healthcare, examining key data types and sources proves critical.

Real-World Data Types

Health data can be pulled from numerous sources to provide valuable insights into the patient journey. 

Claims Data

Claims data results from processing a healthcare claim. Two types of claims data include open and closed. Open claims datasets come from claims clearinghouses or providers’ revenue cycle management systems. They cover a large scale of patient lives, but may not represent complete claims coverage for a given patient. 

Closed claims come from health insurance plans or self-insured employer groups. They tend to cover a smaller scale of patient lives, but represent complete claims coverage for a given patient during the time that patient was on the insurance plan or worked at the employer. Claims data is longitudinal in nature and captures a long period of the patient journey, but it does not have as much depth of clinical detail about a particular medical encounter as other data types. 

Because closed claims datasets are very comprehensive, they prove ideal for health economics and outcomes research (HEOR) that considers a patient’s journey, resource utilization, and the economic burden of their condition. Open claims datasets prove less useful for HEOR, due to their incompleteness for a given patient. Given the large scale of patients covered as well as lower data latency, open claims datasets can prove useful for marketing use cases.

Claims data is even more powerful when used in tandem with clinical metrics such as lab data, EHR data, or patient-reported outcomes. The combination of these data sets can provide deeper insight into symptoms, disease progression, and clinical outcomes.

Laboratory and Genomics Data

Lab testing data proves valuable for a variety of use cases in healthcare analytics from market sizing to monitoring disease progression to finding biomarker signals of patients eligible for certain therapeutics. Lab data can provide a deep point-in-time clinical and biochemistry profile of a patient but isn’t as longitudinal as claims data.

Genomics data is a specialty area of lab testing currently growing in popularity within healthcare analytics given the increase in biomarker-targeted therapeutics. Genomics data proves useful for both clinical development use cases where scientists may employ genomics data to inform biomarker selection or in commercial use cases where genomic results may provide input towards building a predictive model to find patients eligible for a biomarker-targeted therapy.

Pharmacy Data

Pharmacy data provides information about which therapies patients use and how the therapies change over time. Pharmacy data proves extremely useful for specialty drugs, which now account for approximately 75 percent of prescription drugs in development. A network of specialty pharmacies, contracted by pharmaceutical manufacturers, typically distributes specialty drugs. The manufacturer will then aggregate real-world data from specialty pharmacies to understand real-world prescribing, dispensing, and medication adherence patterns.

Electronic Health Records (EHR)

As patients move throughout the health system, valuable real-world data is collected as part of their electronic health records (EHRs). EHR data contains richer clinical detail than claims data, but it may not be as longitudinal because patients could visit many doctors across different care settings over the course of a year, and those doctors may use different EHR systems. EHRs contain data on appointments, medical history, diagnoses, symptoms, medications prescribed, labs, and chart notes. This data is important for gaining a more granular understanding of clinical patient outcomes. 

Even though EHR data is valuable, it requires significant data curation and cleaning because much of the valuable information may reside in unstructured physician notes fields. 

As with all real-world data, biases may exist. For instance, EHR data can have ascertainment bias because the recorded information depends on why the patient came in for a hospital visit, who cared for them, and the intended use of the information. Patients in the emergency room for a broken arm won’t have their insomnia noted in their health records, but if they go to a neurologist, it’s more likely to be noted. 

Generally, the information recorded as part of a patient’s EHR—whether they’re in an inpatient setting, outpatient setting, or a specific therapeutic area—includes:

• Procedures performed
• Diagnosis
• Vital Signs
• Laboratory results
• Medication orders
• Medications administered
• Patient surveys or questionnaires
• Surgical care information
• Symptoms

Note that electronic health records on their own may not contain all of the necessary RWD, so researchers may be required to seek multiple sources of data.

Oncology Data

RWD plays an important role in answering a variety of research questions surrounding cancer. One of the top priorities in research is generating accurate evidence on the efficacy of cancer prevention, diagnosis, and treatment in a real-world setting. 

Researchers often study cancer treatments in a select population in a clinical trial setting. However, researchers can also collect and analyze real-world oncology data to provide RWE on the efficacy and tolerability of new treatment methods in the real world. The main sources of real-world oncology data include:

• Registries
• Claims
• EHRs
• Specialty data providers and networks

Each state legally mandates central cancer registries, thus providing a census of all the patients who have cancer within a defined geographic area. Because of this and the capture of detailed exposure information such as diet or physical activity and patient-reported outcomes, these registries provide unique information because it comes from a non-random group of people. 

Limitations of cancer registry data include a lack of information on outcomes other than survival as well as long-term treatment. Addressing these limitations requires new initiatives such as linking registry data with data from other organizations. The new initiatives, as well as real-time access to pathology reports, provide opportunities to supplement the understanding of therapeutic advances and impact outside of clinical trials.

Consumer Data

Recently, researchers have increased the demand for consumer data. This information can provide additional context about a patient population, such as:

• Jobs
• Socioeconomic status
• Interests
• Health
• Race
• Ethnicity
• Languages spoken

This data comes from consumer data companies and has traditionally been used for targeted marketing. Note that consumer data companies only have data on adult consumers. 

Social Determinants of Health (SDOH)

Social determinants of health (SDOH) are the conditions in which people are born, work, live, play, age, and worship. These have a large impact on peoples’ health, quality of life, and functioning. Some examples of social determinants of health include: 

• Racism
• Polluted water and air
• Access to healthy, nutritious foods
• Physical activity opportunities

Social determinants of health contribute to health inequities and disparities. For example, those who don’t have access to grocery stores that carry healthy foods may not have good nutrition, which can lead to obesity, diabetes, and heart disease. Data on SDOH and various groups of people provide a key understanding of health disparities as well as ways to improve conditions in peoples’ environments. 

Real-World Data Sources

Different types of data providers are relevant for different situations. Three types of data provider categories exist:

Data Platforms

Data platforms provide a technology platform that has intuitive user interfaces (UI) for analyzing data within the platform. These companies have data science and data engineering teams that clean and standardize continuous streams of data coming into the platform, and combined with the UI layer, can be considered user-ready. 

In many cases, the platform provides limited ability to export data for use. Working with a data platform is best for companies without data analytics or data engineering capabilities. 

Data Aggregators

Data aggregators offer cleaned and standardized data that has been aggregated from many underlying sources. Typically, a technology platform or user interface overlay doesn’t exist, and the data is available to license as a one-time or continuous data feed. 

This data is analytics-ready. Companies working with a data aggregator need to have data analytics or business intelligence analysts who can manipulate the data into analysis, but they do not need to have sophisticated data engineering to clean and standardize the data. 

Data Originators

Data originators are closest to the source. They have the most granular and detailed data, but they do not clean it. This data requires the application of sophisticated data engineering capabilities before it can be analytics-ready. 

Real-World Data Solutions Providers

The RWD ecosystem includes both real-world sources, as described above, as well as solutions providers that have built analytic and workflow solutions on top of real-world data. Many platform companies are also solutions companies, having built specific data views and analytic tools that provide solutions for specific use cases. 

Common commercial analytics and clinical development solutions built on top of real-world data include:

Commercial Solutions

• Specialty pharmacy aggregation: These companies aggregate specialty pharmacy data on behalf of pharmaceutical manufacturers to monitor therapy launches. Specialty drug data is proprietary data of pharma companies that may link to other real-world data such as claims for a longitudinal view of the patient journey.
• Outcomes and patient journey: These companies enable outcome studies and patient journey research. Many of these companies build their solution on top of aggregated and linked claims data to enable a comprehensive view of patients as they move through the healthcare system.
• Commercial triggers: These companies provide triggers to commercial teams at pharmaceutical companies to alert them when a patient eligible for a specific therapy sees their provider, so sales teams can be deployed to the provider’s office for education on the relevant disease or therapy. Use of this solution is especially common in rare diseases since providers are often unaware of the rare disease and patients can be hard to diagnose.
• Digital marketing: These companies identify relevant patients and providers and then serve up digital advertising to educate them on a disease or therapy.
• Commercial analytics and insights: These companies provide aggregated data, usually claims or EHR, to help commercial teams with strategy and insights before and post-launch.

Clinical Solutions

• Trial recruitment: These companies use aggregated data—typically EHR, lab, and claims data—to identify the ideal clinical trial sites that have sizable populations of patients who would meet inclusion/exclusion criteria for a trial.
• Synthetic control arms: Synthetic control arms, also known as external control arms, are where real-world data is utilized as the control arm rather than enrolling actual patients into a control arm where a placebo or standard of care (SOC) is utilized. 

This is popular in disease states where patient populations are

increasingly sub-stratified by biomarker status (e.g., oncology, rare disease) given the challenges of recruiting enough patients and ethical considerations of placing patients on placebo or standard of care (SOC). Companies that provide these solutions often have deep and highly curated clinical and genomic data to conduct synthetic control arms. Synthetic control arms lower trial costs, increase efficiency, and increase the speed of therapies to market. 

Decentralized clinical trials: These companies provide technology infrastructure to collect data and support decentralized clinical trials (DCTs). DCTs are trials where patient communication and data collection has been decentralized away from a traditional clinical trial site. Instead, remote and digital technologies communicate with study participants and collect their data.