Systems scientists have an adage that “everyone is right, but only partially so,” meaning that every person brings a unique perspective to a situation, but no one person can have the whole picture. This is true in health data systems as well. We have made great strides in data aggregation within health information exchanges (a tremendous benefit to personal health) and all-payer claims databases (a tremendous benefit to population health and health policy), but still, no single data source holds the whole health story both of an individual and a population. This distinction in use case(s) is important.
To benefit both personal and population health, we must integrate data across sources, time and geographies in a way that makes it easy to contribute data and easy to get valuable information and analytics in return. There are several important components and benefits of this model to consider.
Data governance to support integrated data — and benefit both population and personal health — is critical. This integrated data system will be most successful if it is developed by and operated for the good of the community, built on a common agreement and trusting relationships.
Interoperability, with standardization wherever possible, prevents silos of data or information and ensures meaningful combination of data across sources, time and geographies. Shared services, meaning that the same resources and technologies are deployed on all integrated data, are critical to ensuring improvement in both health and health care at an individual and population health level. Shared services need transparent methodologies to address data integration and normalization, identity resolution and provider attribution; to involve patients; and ultimately to share actionable results through advanced analytics and reporting.
Data quality for personal health use cases differs from data quality requirements for population health use cases. Any use case for patient care requires a much higher level of accuracy than use cases for population health. The acceptable quality thresholds for timeliness, completeness and accuracy will depend on the use cases. But the threshold for acceptable data quality requires human judgement and ideally a trusting group of partners to agree on what that threshold should be.
Initiatives to improve health equity and align measure sets can benefit greatly from integrated data. With information from both claims and clinical data, collaborating partners can increase standardization of measurement. Rather than running separate measure sets with different data sources, they can align measure sets and measurement sources. Greater alignment can reduce administrative burden for providers, as well as increase efficiency in quality improvement and alternative payment methodologies.
Finally, integrated data increases the power of health data to drive descriptive, predictive and prescriptive analytics. Descriptive analytics use population-based, historical data to describe what has already happened within a population and are useful for performance measurement, research and program evaluation. They primarily benefit population health. Predictive analytics predict what could happen for an individual or a population, and are useful for risk scores and risk models, epidemiological surveillance, public health prevention programs. More than that, predictive analytics benefit both patient care and population health. Prescriptive analytics can prescribe care for an individual and are primarily useful for clinical decision support.
It is time to look across the health system and across data sources, and plan ahead for a data ecosystem with integrated claims, clinical and other data sources that can benefit both patient care and population health use cases. The value of integrated data is greater than the sum of its parts.
Learn More
Learn about Comagine Health's data solutions services or contact us with your questions about innovative, effective data integration.