2021 Health Data Review — Part I: Top Trends in Health Data Demand
Introduction
As Head of Data Strategy at Datavant, I help clients find partners with relevant real-world data (RWD) for linkage to their own data or to create multi-partner linked datasets. These clients include pharmaceutical companies, health systems, government, payers, and analytics platforms. Connecting data helps them create a robust picture of patient health to accelerate research, improve care, and lower healthcare costs. Clients come with questions like:
- How can I find patients with fatty liver disease for a clinical trial?
- Is this new migraine therapy cost-effective?
- What factors are preventing patients from coming into the clinic for follow-up care?
- What are the risk factors contributing to this patient’s high healthcare costs?
I’ve heard hundreds of client questions and learned a lot about our partners’ datasets. I’d like to share my experience and help others find data needed to improve patient health.
Healthcare is the Biggest Producer of Data
New health policies, technologies, and scientific discoveries have resulted in a massive increase in the volume and variety of health data. The 2009 HITECH Act accelerated the use of electronic health records (“EHRs”). The decreased cost of genomic sequencing and increase in biomarker-specific drugs has led to growth in genetic testing.1 Smartwatches, wearables, and health apps are now commonplace.2 The pandemic accelerated the use of sensors and remote technologies. Cloud and data science enable storage and analysis of data at scale. RBC Capital Markets estimates that healthcare generates the largest volume of data. At a 36% compound annual growth rate by 2025, healthcare data is growing faster than data in any other industry.3
New Analytics Companies Are Formed Each Year
With more health data, new analytics companies have sprung up to analyze it. From patient recruitment and clinical trial technologies to population health and value-based care analytics, thousands of organizations serve biopharma, device makers, payers, patients and providers. As a result, Datavant now connects the largest health data ecosystem with thousands of data originators, data recipients and data platforms. As we head into 2022, here are the five most common types of data our clients sought out last year and the questions they were trying to answer.
Demand Trend #1: Demographics and Social Determinants of Health (SDOH)
As COVID entered its second year, the problem of health inequity in America became salient.4 Many groups such as people of color, women and other vulnerable populations are neither adequately represented in research nor have equal access to quality care.5 Government, life sciences, health systems and payers all want to understand the factors associated with worse health outcomes in these groups to try and close these gaps.
Demographic variables include inherent characteristics in people (e.g., age, race, gender, and marital or family status), whereas SDOH are the conditions under which people live, work, and learn. (e.g., education level and employment status). Datavant has numerous partners with demographic and SDOH data. The utility of specific SDOH attributes depends on the question a client is trying to answer. I’ll often work with clients to understand their health-related questions and help them find the best SDOH data variables to answer them.
Demand Trend #2: COVID Vaccination & Variant Data
Since the pandemic, data on COVID diagnoses, vaccination rates and variants have been in high demand and we’ve seen an uptick with the Omicron variant. Unfortunately, this data remains fragmented given the state-by-state approach to reporting COVID cases. Some of our largest data aggregator partners have vaccination status on just ~20% of all 200mm vaccinated individuals in the US, so linking data for a national view of U.S. patients is critical. Datavant lends its linking technology to support the NIH’s National COVID Cohort Collaborative (N3C). Our connectivity technology also supports the national COVID-19 Research Database (CRDB), which is used by academic researchers and was recently recognized by the Reagan Udall Foundation with an award for Innovations in Science. If you are interested in the application process to use the CRDB you can find more details here.
Demand Trend #3: Oncology Patient Journey
Linking data to understand the full oncology patient journey is the most frequent type of linkage we see. The oncology patient journey is complex. Patients may present at a primary care office and be referred to an oncologist who orders labs such as a biopsy, diagnostic imaging or genetic testing. They may visit radiologists, undergo surgery, and take drug regimens that include infused and oral drugs. While survival is a key outcome in oncology studies, mortality data is not always captured in EHR datasets.6
Understanding the oncology patient journey often requires linking claims data including medical, retail pharmacy and specialty pharmacy claims, to EHR data from ambulatory care, integrated delivery networks, and community oncology clinics. In particular, clients struggle to find oncology EHR data from academic medical centers, which are hospital systems that are affiliated with a research university. Academic medical centers (AMCs) are very important care settings that conduct research, provide education, and deliver clinical care. They have access to cutting-edge technologies and can serve patients with rare cancers by offering advanced procedures like bone marrow transplants and novel drug trials. Clients seek pathology, radiology, labs, imaging and genetic data which necessitates linking data from multiple partners specializing in cancer. In addition to mortality data that covers more than 85% of weekly death events, the Datavant ecosystem includes many oncology-specific data partners:
- 5 community-focused oncology RWD providers,
- 3 AMC-focused oncology RWD providers,
- 8 genomic data providers (3 of whom also have germline testing data),
- 9 providers of ambulatory EHR data, and
- 6 of the top 10 AMCs in cancer.
Demand Trend #4: Specialty Drug Data
Specialty drugs dominate pharmaceutical company pipelines.7 These drugs are distributed by specialty pharmacies (SPs). Many SPs only provide usage data on that drug back to the drug manufacturer for Limited Distribution Drugs (“LDDs”). In such cases, the drug and often the entire drug class is not available to aggregators and represents a gap in commercial databases. To compensate, life science companies connect their 1st party SP data feeds to 3rd party datasets for a more complete picture of patients’ adherence to therapy and their drug’s utilization versus the competition.
Demand Trend #5: Rare Disease Patient Data
Rare diseases affect fewer than 200,000 people in the U.S.8 Rare disease patients often endure a diagnostic odyssey that runs 5–7 years. They may see more than 7 specialists before being properly diagnosed which creates a highly fragmented data journey.9 Lastly, drugs for these patients are primarily specialty pharmacy-distributed, so data is often unavailable within commercial datasets. Clients seek our help linking data across many partners to find patients for clinical trials and observational studies.
Demand Trend #6: Linking First-Party Data
Biopharma clients have begun linking internal 1st party data to external 3rd party data assets they license. First-party data includes clinical trial data, specialty pharmacy data, hub services data, patient registries, sponsored testing data, digital engagement data, and other assets. Linking 1st party and 3rd party data creates a comprehensive view of patients’ health. For example, connecting data from hub services to Rx claims can shed light on the time between a patient’s attempt to get on therapy to script fulfillment and ongoing adherence. Connecting clinical trial data to EHR and claims datasets extends data collected for long-term outcomes analyses. There has been significant demand in linking these pharma-proprietary data assets to create more value and insights from enterprise data.
Demand Trend #7: Enterprise Data Linking
One of the most exciting trends we are seeing with biopharma clients is an enterprise-level data strategy that includes tokenizing every trial and health economics and outcomes study (tokenizing refers to de-identification of patient identifying information (PII) and assignment of a hashed, encrypted token to represent the patient). Centralized evidence generation teams are using the approach to maximize external partnerships as a data partner may have data that matches patients across several studies. Tokenizing all studies enables faster partner identification and negotiation of data for multiple studies at one time.
Summary: Implications for Data Users, Data Originators and Platforms
This year we expect to see growth in use cases that require granular clinical, genetic, pathologic, imaging, and biomarker features. Those seeking data need to define clear criteria for the population of interest, the questions they’re trying to answer, and the use case. Preparing these in advance will accelerate finding a partner with the best data to meet those needs.
For data originators, aggregators and analytics providers, understanding demand-side trends can help them maximize the relevance of their data to answer key questions. Making tokenized data available for data users to run on-demand overlaps enables rapid identification of relevant partners that have shared patients. In early partnership diligence conversations, sharing information about data content, quality and curation advances partnerships quickly.
At Datavant, we believe that every decision in healthcare should be powered by complete patient data, and the more we can help partners find each other, the more we enable everyone’s effort to improve patient outcomes. I’ll be back soon with an update on trends in new data types (supply-side trends). In the meantime, feel free to reach out to me and provide feedback at su@datavant.com.
Thank you to Elenee Argentinis, Head of Marketing @ Datavant for help in editing this article!
References:
- The Cost of Sequencing a Human Genome. The National Human Genome Research Institute. Published November 1, 2021. Accessed January 5, 2022.
- How Fitbit, Whoop and Other Gadgets Are Measuring Brain Activity, Glucose and Sleep. Betsy Morris, The Wall Street Journal. Published April 12, 2021. Accessed January 5, 2022.
- The Healthcare Data Explosion. RBC Capital Markets. Accessed January 5, 2022.
- Health Disparities and the Coronavirus Disease 2019 (COVID-19) Pandemic in the USA. Khatana SA, Groeneveld PW, Journal of General Internal Medicine. Published May 27, 2020.
- Representation in Clinical Trials: A Review on Reaching Underrepresented Populations in Research. Yates I, Byrne J., Clinical Researcher. 2020;34(7). Accessed January 6, 2022.
- Validation of a Mortality Composite Score in the Real-World Setting: Overcoming Source-Specific Disparities and Biases. Michelle H. Lerman, Benjamin Holmes, Daniel St Hilaire, Mary Tran, Matthew Rioth, Vinod Subramanian, Alissa M. Winzeler, and Thomas Brown. JCO Clinical Cancer Informatics 2021:5, 401–413.
- A look at Specialty Pharmacy Dynamics. Shah P, CVS Health. Published July 13, 2020. Accessed January 5, 2022.
- How Many Rare Diseases Are There? Haendel M, Vasilevsky N, Bologa C, Harris N., Nature Reviews Drug Discovery; 2019.
- The Diagnostic Journey For Rare Disease Patients. Avalere. Published June 2021. Accessed January 5, 2022.