Analytics in Healthcare

An emerging trend in healthcare reflects the new strata of data management and analysis expanding across other industries. The 2001 founding of the Healthcare Data Warehousing Association (HDWA), now with a professional membership of more than 300 organizations throughout the US and Canada, has effectively advanced the role of analytics in healthcare. The trend is to progress through several identifiable phases of data management.

Phase 1 — Data Collection: characterized by the broad adoption of electronic medical record (EMR).

Phase 2 — Data Sharing: characterized by expansion of health information exchanges (HIEs).

Phase 3 — Data Analytics: characterized by detectable patterns across aggregated data becoming useful, and by the development of enterprise data warehouses (EDW), i.e., big data utilization.

As exemplified by organizations such as Intermountain Healthcare (using HELP EMR) and the US Veterans Affairs (VA) Health Care system (using Vista), integrated care models in the US and elsewhere have demonstrated the significant impact of leveraging big data. For example, the VA’s health system outperforms private sector health care systems in:

  • adhering to recommendations for patient care processes.
  • following clinical guidelines.
  • higher rates of drug therapies that are evidence-based.

These VA achievements are largely due to EMRs and analytics that frequently allow VA professionals to effect a closed loop on clinical practices, thereby enabling the organization’s practices of performance-based disease management and framework for accountability.

Why Embark on Healthcare Analytics?

Healthcare in the US and other countries has gradually been progressing through the three basic phases of data management mentioned above. The healthcare industry is increasingly developing these basic capabilities for a system-wide elevation of quality and reduction of cost—the benefits of a truly data-driven healthcare culture. Methodically building on foundational components of the model is necessary in order to establish the robust support required for building an ultimately sustainable and impactful data management system.

Levels of the Healthcare Analytics Adoption Model (HAAM)

The Analytics Adoption Model (AAM) affords the benefits of structuring per a standard educational curriculum and is based on long-term healthcare industry observation and learning. On each of eight levels of incorporation, an organization expands its analytic capability in new data sources, complexity, data literacy, and data timeliness. At each new level, data binding develops in complexity.

Level 0 Fragmented Point Solutions: At this base level, organizations’ analytics capabilities are typically limited to departmental analytics in finance, pharmacy, laboratory, acute care nursing, and physician productivity. Applications are not integrated. So, new knowledge is isolated, and sub-processes may be optimized at the greater expense of organization-wide processes. Data content overlaps lead to discrepancies in versions of the truth. Generating reports is typically inconsistent and labor intensive. And, no formal function exists for maximizing data quality and value.

Level 1 Enterprise Data Warehouse: The organization’s sources of HIMSS EMR clinical data, patient experience data, materials and supplies data, and financial data (particularly costing data), are located together in the same data warehouse (local or hosted). And, across the enterprise, a searchable repository of metadata is available.

Level 2 Standardized Vocabulary Patient Registries: Core data content is related and organized. Master reference data and vocabularies are identified and established as the standard across the EDW’s disparate content of source system data.

Level 3 Automated Internal Reporting: Internal reporting is automated. Production of reports is more efficient, accurate, and consistent. Maintenance of reports involves no labor or merely small amounts of labor to support. Reports are obtained virtually entirely by self-service for the organization’s board and executive level operations and management to use in generating analytics for key performance indicators. An analytics user services group assures data quality, trains personnel for data literacy and guides strategy for mission-critical data acquisition and further levels of adoption.

Level 4 External Reporting: There is adaptability to shifting requirements and agility in report production. The EDW’s data content has been broadened to include text data clinical reports and notes and from patient records. Text query tools, using basic keyword searches in patient records, are based on the EDW.

Level 5 Waste & Care Variability Reduction: Variability across care processes is reduced. The organization can be differentiated in the market by a significant quality and cost distinctions enabled by its healthcare data analytics. Data is used explicitly for informing healthcare strategies and formulating policy. Measurement of conformance with clinical best practices is facilitated, as well as minimization of waste. Improvements in individual patient care are suggested through population-based analytics. Clinical and cost data, as well as patients’ information regarding insurance claims, are combined in the EDW content.

Level 6 Population Health Management Suggestive Analytics™: A sustainable healthcare environment in which clinical outcomes are more thoroughly understood through analytics has been established. At the point of care, analytics are available to support optimization of patient care quality, population management, and healthcare economics. The data content of the EDW has been expanded to include patients’ external pharmacy data, bedside devices, home monitoring data, and other detailed activity-based costing. The organization realizes financial rewards associated with clinical outcomes.

Level 7 Clinical Risk Intervention & Predictive Analytics: Focus can expand from case management to collaboration between clinicians and payer partners to include predictive modeling and risk stratification. Hospitals, physicians, payers, employers, and patients collaborate and share risks and rewards (for example, financial rewards of a patient’s healthy behaviors). Provisions are made in the system for patients not participating in healthcare protocols, due to cognitive, geographic, or other constraints to bring them into the system. Data content is widened to include home monitoring and long-term care facility data, as well as patients’ reports of outcomes. EDW updates are executed within an hour from the time of changes to the source system (on average).

Level 8 Personalized Medicine & Prescriptive Analytics: At the point of care, prescriptive analytics based on populations’ outcomes are available to improve an individual patient’s outcome. Data content is expanded to include around-the-clock availability and usability of biometric data, genomic data, familial data, and patient environment data, seven days per week. EDW updates are executed within minutes of a change to source systems. New data warehouse content is combined with algorithms (to be discovered) which can identify relationships between data of these various types.

When Data Analytics Become Impactful

Deployment of EMRs (characterizing the data collection phase) cannot have a significant impact on healthcare quality or cost. (And, HIEs have proven to be an unsustainable economic model.) Nevertheless, investment in EMRs is fundamental to realizing the value of healthcare analytics. The tools and outcomes that have lent to the adoption of EMRs (especially the EMRAM) have led the HAAM to pursue a framework for quicker progress toward analytic maturity.

Emerging technologies will further permit treatment tailored to individual patients based on insights acquired from new data sources and algorithms. Boundaries of medicine are extended to include evidence from EDWs shared with clinical content providers, to enhance knowledge content from randomized clinical trials.

Purpose and Benefits of the HAAM

The Healthcare Analytics Adoption Model provides:

  • a framework within which organizational leaders can evaluate industry adoption of analytics.
  • a well-mapped process that organizations can follow along to evaluate their own progress toward full adoption of analytics-based healthcare management.
  • a framework within which vendor products can be efficiently evaluated.

This model helps healthcare provider organizations to understand thoroughly and to leverage analytics capabilities fully, to achieve the elusive goals of improving quality of care and lowering costs.

Self-Assessment Survey

As with the HIMSS EHR Adoption Model, a logical progression is necessary to becoming a systemically analytics-driven healthcare organization. Health systems often become frustrated in their scattered efforts to grapple with analytics, when inadequate tools, platforms, and skills are not mastered at appropriate levels of the analytics adoption process. The self-assessment survey is available to assist you in evaluating your organization’s current level of functionality in healthcare data analytics. A customized summary of your survey results (in HTML format or PDF version), along with a list of input-based recommendations is then generated, which you can email to yourself.

Health Catalyst

We help healthcare institutions develop data analytics in healthcare systems toward the realization of return on investment of EMRs, developing an industry practice of data warehousing, and building a modernized data-driven healthcare industry culture that is economically incented to promote optimum health at the lowest possible cost.

To whatever level a health system has progressed in its analytics, Health Catalyst develops and deploys solutions to accomplish specific clinical and operational outcomes in data systems that can deliver valuable patient care insights, identify inefficiency and waste, and monitor bottom-line impacts of changes and improvements to healthcare system processes. Our areas of expertise include the following, among others:

  • care management
  • research informatics
  • clinical analytics
  • operational management
  • performance management
  • financial decision support
  • population health and accountable care

For More Information

If you are interested in obtaining a deeper understanding of the eight successive levels of healthcare data analytics, see the white paper explaining the HAAM (by Dale Sanders, Dennis Protti, and others). Additionally, a live webinar detailing the model, as well as slides, and a transcript of the webinar are available.