Domain

Learning Health System

Type

Case Study or Comparative Case Study

Theme

population; quality; operations

Start Date

7-6-2014 2:55 PM

End Date

7-6-2014 4:15 PM

Structured Abstract

Introduction:The published care coordination literature has repeatedly shown that achieving net cost-savings requires careful targeting of expensive clinical resources to those most likely to have future hospital admissions. However, identification of high-risk populations requires integrating several different types of current and historical data and making it available nearly real-time at the point-of-care.

Background/Context:The Center for Medicare and Medicaid Innovation (CMMI) awarded Denver Health’s (DH) integrated, safety net health care system $19.8 million to transform its primary care delivery system to provide individualized and enhanced clinical care and health information technology services (HIT) to a risk-stratified population to meet the Triple Aim of improved health and patient experience while reducing costs.

Innovation Leveraging DH’s broad data capture as an integrated delivery system with its own health plan, DH combined individual-level data from multiple clinical and administrative sources (e.g., registration/billing/health plan data) to facilitate patient identification, program design and implementation. Specifically, DH applied evidence-based methods to group approximately 60,000 current adult patients into clinical and financial risk strata according to: diagnosis, clinical registry status, past utilization, financial information, demographic characteristics, and predictive modeling tools. In addition, historical data tables were created for analysis purposes by repeatedly applying the risk stratification algorithm to produce monthly cross-sectional snapshots of approximately 60,000 each of targeted adult patients for the period between 5/1/2011-present. Using BI tools, DH conducted numerous simulations on the historical data file to optimize the risk stratification algorithm, target interventions to populations most likely to benefit, and produce clinical staff worklists useable at the point of service. For example, diabetes clinical registry information (lab results) was integrated with predictive modeling results, and DH-specific, population-level historical utilization information to target patients for clinical pharmacist services. The historical data tables were particularly useful in identifying patient populations that had a history of high utilization that was stable or increasing over time.

Lessons Learned:

  • Identification of high risk populations is complicated by regression to the mean of currently high-utilizing populations
  • Population-level historical utilization patterns can augment current year claims trends and predictive modeling tools to ID populations most likely to benefit from care management resources due to a pattern of consistently high spending.
  • Dynamic, business intelligence tools can facilitate interdisciplinary program design decisions by enabling real-time changes in data views (e.g., drilling down from population-based to a patient-specific).

Acknowledgements

•This presentation was made possible by Grant Number 1C1CMS331064 from the Department of Health and Human Services, Centers for Medicare & Medicaid Services.

•The contents are solely the responsibility of the authors and have not been approved by the Department of Health and Human Services, Centers for Medicare & Medicaid Services.

Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 License.

Share

COinS
 
Jun 7th, 2:55 PM Jun 7th, 4:15 PM

Integrating Clinical, Billing and Demographic Data for Care Coordination Program Design and Implementation

Introduction:The published care coordination literature has repeatedly shown that achieving net cost-savings requires careful targeting of expensive clinical resources to those most likely to have future hospital admissions. However, identification of high-risk populations requires integrating several different types of current and historical data and making it available nearly real-time at the point-of-care.

Background/Context:The Center for Medicare and Medicaid Innovation (CMMI) awarded Denver Health’s (DH) integrated, safety net health care system $19.8 million to transform its primary care delivery system to provide individualized and enhanced clinical care and health information technology services (HIT) to a risk-stratified population to meet the Triple Aim of improved health and patient experience while reducing costs.

Innovation Leveraging DH’s broad data capture as an integrated delivery system with its own health plan, DH combined individual-level data from multiple clinical and administrative sources (e.g., registration/billing/health plan data) to facilitate patient identification, program design and implementation. Specifically, DH applied evidence-based methods to group approximately 60,000 current adult patients into clinical and financial risk strata according to: diagnosis, clinical registry status, past utilization, financial information, demographic characteristics, and predictive modeling tools. In addition, historical data tables were created for analysis purposes by repeatedly applying the risk stratification algorithm to produce monthly cross-sectional snapshots of approximately 60,000 each of targeted adult patients for the period between 5/1/2011-present. Using BI tools, DH conducted numerous simulations on the historical data file to optimize the risk stratification algorithm, target interventions to populations most likely to benefit, and produce clinical staff worklists useable at the point of service. For example, diabetes clinical registry information (lab results) was integrated with predictive modeling results, and DH-specific, population-level historical utilization information to target patients for clinical pharmacist services. The historical data tables were particularly useful in identifying patient populations that had a history of high utilization that was stable or increasing over time.

Lessons Learned:

  • Identification of high risk populations is complicated by regression to the mean of currently high-utilizing populations
  • Population-level historical utilization patterns can augment current year claims trends and predictive modeling tools to ID populations most likely to benefit from care management resources due to a pattern of consistently high spending.
  • Dynamic, business intelligence tools can facilitate interdisciplinary program design decisions by enabling real-time changes in data views (e.g., drilling down from population-based to a patient-specific).