Healthcare business intelligence
Download as PPTX, PDF6 likes5,975 views
The document discusses the value of analytics in healthcare. It argues that healthcare organizations have large amounts of data from various sources like EMRs, finance systems, and other clinical and administrative systems. However, this data is often underutilized. The author proposes using a healthcare enterprise intelligence framework to extract, transform, and load this data into a centralized data warehouse where it can be integrated, standardized, and made available for analysis. This would allow healthcare leaders to better understand their operations and make more informed decisions using business intelligence tools like OLAP cubes, dashboards, and reports. The goal is to improve outcomes by personalizing care based on past patient data and evidence.
Healthcare business intelligence
- 1. HEALTHCARE BUSINESS INTELLIGENCE AND ANALYTICS AN EXPLORATORY EVALUATION AND ASSESSMENT ON THE VALUE OF ANALYTICS IN HEALTHCARE Nick Sullivan, MHA
- 2. ABOUT ME • UNC Grad („07 B.A Public Policy, „12 Masters Healthcare Administration) • Gained interest in analytics while working as grad assistant at School of Public Health Business & Finance office • Currently employed as Administrative Fellow at Novant Health in Charlotte, NC • Looking to advance knowledge technical, clinical, strategic and financial aspects of healthcare business intelligence • San Francisco 49er Fan
- 3. ABOUT MY EXPERIENCE • Healthcare is a mammoth industry • Healthcare is undergoing sweeping and disruptive changes • Healthcare leaders are inundated with multiple competing and uncertain priorities • Healthcare organizations must learn to do more with less • … • Healthcare will get better.
- 4. SHOPPING FOR HOLIDAY GIFTS A LOT LIKE HEALTHCARE? • What am I going to get for • What information could I use my family this year? about my family to make a better decision? • 2 parents, 4 sisters, 4 nephews, 6 neices, 2 brother in-laws • Age • How much am I willing to • Gender spend in total? • Needs • How old are my nieces and • Wants nephews? • Interests • Does that affect whether or not I get them clothes or • Current Trends toys? • Satisfaction with previous • Can I get them the same gifts gifts and not feel bad about • Frequency of use of previous it? gifts • Did they like my gift from last year? Questions + (∑Facts/Beliefs) – Noise = Knowledge Better Decisions Better Outcomes
- 5. RELEVANCE TO HEALTHCARE • Grown Accustom to shaping processes and decisions based on: • intuition, • provider preference, • amount and type of resources available, • competing priorities, • vested financial interest and • incentives aimed at more care is better care (do it all) • What if we had information to make decisions based on individual patient characteristics and evidence gleaned from previous encounters with the disease? • How do we provide timely, efficient and cost-effective care that resulted in ultimate patient satisfaction? ANALYTICS
- 6. THE RISING TIDE OF DATA • World becoming awash in data, growth at 60% annually • Widespread Healthcare EMR implementation will rapidly expand access to data • How does healthcare make the most of its growing data?
- 7. THE VALUE-ADD OF ANALYTICS • Healthcare Organizations must find ways to converge different types of data to glean insight on critical aspects of running the enterprise: Clinical Administrative Financial Operational • But we already create departmental reports, correct? Analytics v. Reporting Business Intelligence Area Reporting Analytics Analyst Primary Function Building Questioning Use of Visuals Configuring Examining Data Relationships Consolidating Interpreting Data Sourcing Collecting Connecting Data End Game Summarizing Validating Communication Method “Push” “Pull” Data Lifespan Static Dynamic Data Orientation Look Back Look Ahead
- 8. THE HEALTHCARE ENTERPRISE INTELLIGENCE FRAMEWORK Source Data Staging Data Warehouse Customization Client Ad Hoc Query Practice Finance EMR Mgmt. Service Line Lab Pharmacy Transform Disease Specific HR Payroll Clean Condition KPI‟s Fully Integrated Extract Scrub Standardized Scorecards, Reports, Surgery Dept. Merge Load Patient Dashboards Validate Historical Centers Sprdshts Registries Confirm One Version of Truth Anomaly Detect Secure Strategic Legacy Planning Physician Mapping Clinical Clinic Service Sys Line Graphs & Charts Patient Scheduling Satisfaction Errors Costing, Finance Multidimensional Data Mining Market Reg. & P4P Operating Data Reqmts. Room Metadata
- 9. KEY ENTERPRISE ELEMENTS Source Data: data that is critical to running the business. - Typically operational in nature and built to handle large numbers of simple, predefined read/write transactions using OLTP - Integrated into data warehouse for analytical use (OLAP) Focus Area Operational System Data Warehouse (OLAP) (OLTP) Orientation Application Oriented Subject Oriented Business Use Used to run business Used to analyze and optimize business Data Presentation Detailed & Discrete Summarized and refined Time Orientation Current, Up to Date Snapshot of Data Data Relationships Isolated Integrated Frequency of Use Repetitive Access Ad-hoc access Primary User Business Processer Business Analyst
- 10. EXTRACT, TRANSFORM, LOAD (ETL) ETL: Process of gathering, preparing and integrating data into the data warehouse Extraction: data taken in “as-is” format from source Transform: data cleaned, validated and confirmed for eligibility for inclusion into data warehouse Load: maps source data attributes to schema of data warehouse Most critical part of data warehousing process as this defines, creates and maintains the integrity of the enterprise data.
- 11. DATA WAREHOUSE Repository for organizational data, ultimate source for reporting and analysis: Subject-oriented • The data in the data warehouse is organized so that all the data elements relating to the same real- world event or object are linked together. Non-volatile • Data in the warehouse is never deleted or replaced. Once the data is in the data warehouse, it is permanent and kept for reporting purposes. Integrated • Contains data from nearly all of the organizations operational systems. Time-variant • Contains a component of time for every operational data element.
- 12. CUSTOMIZATION Datamarts: subsets of data warehouses that contain a much smaller set of data typically focusing on one business area. • quicker access to specific information that certain groups • Dependent on data warehouse, does not interfere with integrity • Gives “ownership” to individual business units over specific data • Allows business units to create and track metrics, targets, KPI‟s and performance goals
- 13. CUSTOMIZATION • Online Analytical Processing (OLAP): software process that provides a multidimensional view of enterprise data. • Fast • Consistent • Iterative process • Reflects familiarity with user understanding of business • Uses data cubes to create multidimensional views
- 14. OLAP CUBE Provides users the ability to create relationships and multidimensional views of different data sources 33 71 AMI 22 1 Can perform functions such as: 1 12 61 1 Disease CHF • slicing • dicing COPD 54 10 15 81 • pivoting • rolling up Pneum 42 122 132 19 11 • drilling down A B C D Physician
- 15. CLIENT: REPORTING AND ANALYSIS Ad Hoc Query: Highest level of client customization. Gives user liberty within certain constraints to work directly with raw data Level of Data Granularity Multidimensional Data Mining: Use of OLAP tool and cube to create various views Scorecards, Dashboards Reports: pre-defined views and KPI‟s for specific business units and/or goals. May allow drill down or roll up function Graphs & Charts: typical visual representation of predefined metrics and views
- 16. ANALYTICS AND HEALTH REFORM Patient Protection and Affordability of Care Act - Signed into law 2010 Focuses on Triple Aim of: Increased Access, Improved Quality, Cost Reduction OLD: NEW: Fee for Service Access Quality Cost Value Based Care Emphasizes Value-Driven Care and shift from fee-for-service
- 17. ANALYTICS AND HEALTHCARE • Healthcare analytics is intended to improve decision making. Healthcare Decisions can be broken into: Tactical, Operational, Strategic Purpose and Goal Types of Measures Analytical Uses Patient Satisfaction Disease Mgmt. Protocol Adherence Order Set Compliance Episode Profiling Patient Level Tactical Decisions Medication Errors Risk Scoring Provider Performance Activity Based Costing Care Process Variance Process Mapping Care Process Supply Use Value-Add Analysis Operational Stewardship & Cost Process Based Costing Care Coordination Management Gap Identification MD Network Analysis Staffing Predictions Price Setting Pattern and Trend Recognition Strategic Planning & Growth Utilization Predictions Agile Marketing Resource Channeling Community Needs Assessment
- 18. DRIVING VALUE • As reimbursement models change, focus will shift from volume to value and delivering on outcomes. Value = Quality/Cost Value Based 2014 Reimbursement Model: Purchasing Healthcare providers must use data to measure, track and improve performance in these areas.
- 19. USING DATA TO CREATE VALUE Place analytical focus on three aspects of care: Process, Cost and Outcomes
- 20. Care process and improvement feedback loop
- 21. STANDARDIZATION, VARIATION & WASTE • Standardization: applying uniformity across the enterprise throughout every element of care to increase likelihood of desired outcome. • Use data to determine which elements to standardize • Order sets What works best and • Treatment regimens produces the best • Supplies outcomes? Let data tell • Care channeling you, standardize and deploy. • Disease Management Techniques • Variation: deviation from standardized processes • Helps control costs and identify areas for improvement
- 22. WASTE • By standardizing care processes, and applying analytics, variation is spotted and waste or non- value adding elements are discovered. • Equates to a resource that has not yet been discovered or exploited for its value. • Increases capacity to perform primary business functions • Saves time by omitting non-value adding steps • Decreases cost of providing care
- 23. DRILLING DOWN TO REDUCE COSTS • Healthcare providers must deliver on the cost element of the Value = Quality/Cost equation. • Data and drill-down analytics helps remove unnecessary costs • Processes can be analyzed at different levels: • Organization (All diabetes patients) • Population (Females, age 32 -45) • Patient (Ms. Jones) 2 Annual ED visit most visits on common average Young 15Annual Well-visit most visits on All Sickle common average Cell Patients Old
- 24. COSTS OF: REPORTING, TIME TO ACTION & HIDDEN INSIGHT Analytics helps reduce cost by: Productivity cost incurred to 1. Reducing reporting costs. perform activity 2. Increasing “time to action”. Time to reach decision, “action” 3. Freeing hidden insight. Insight gained from business user having access to analytics Business Leader/Analyst has goal/question in mind Business Decision is Business Decision is made made Contacts data owner Owner queues request Data Analyst validates, Analyst presents to Business Leader aggregates, integrates, Information to Business Evaluates Strategies for models, data Leader Solving Problems
- 25. COMPETING WITH ANALYTICS Healthcare is no longer “build and they shall come” - resources have tightened - patients consumers have choice • Using data to enhance reputation and recognition • Appeal to customers with and ability to deliver on promises and showcase facts • Provide patients with customized, patient centered care using data at fingertips.
- 26. SELF-SERVICE BUSINESS INTELLIGENCE No question is a bad question. Putting the power of analytics at the fingertips of business experts and enabling them to question the data 1. Who is effectively managed? Why? (Age, Zip, Ethnicity, Payor, Gender) 2. Who‟s not, why? (seasonality, facility, comorbidities, procedure, visit frequency, appropriate care relationships, age) 3. What is their average total cost, LOS , and #of tests/visit? 4. Did they acquire any infections? If so What kind? 6. Of those not managed, have they had ED Visits? How Many? Time between visits? Did they get better or worse post ED? 7. Who developed post care complications? Why? (procedure error, wrong test, wrong drug, staff competence, infection) 8. What kind of complications where they? 9. Who was readmitted to hospitals? 10. What was their reason for admission? 11. Did they have intermittent communication with provider? If so, who, what type, how many? 12. What do MD, RN Manager notes say about the patients? Any pattern amongst groups? 13. Were they all from same facility? 14. Were they all from same facility?
- 27. COMPETING WITH ANALYTICS EXAMPLE Question: Why is the Cardiovascular Service Line losing market share to the competitor? Data Request: please provide report that shows market share by: With Analytics: VP Drill down capability 50 – 64Pt. Scheduling Cohort 50% Drop in Cases Age System shows Fewer 50-64 age Ethnicity patients scheduled ? Product Increase # of nurse and brand awareness is low Zip Code practitioners to improve throughput Practice Mgr.: MD’s backlogged due to Payor Group elderly throughput Begin Service Type billboard campaign 65+ is directly to market Problem is not correlated with to seniors awareness but cardiovascular ACCESS demand
- 28. Disease "Hotspotting" Readily identifies patients with specific diseases. Alerts, Notifications •Allows for identifying patients with high cost diseases or patients with potential to worsen due to the presence and Decision of a combination of predefined factors. Alert triggers action to monitor patients with targeted follow-up and intervention strategies. Support Systems Gap Identification Tracks whether patients received a service or not in a proces of care. •Removes "chance" from care regimen by hardwiring specific events into process, alerting when a gap is Speeding up the present. Patients can be auto-populated onto a list for specific follow-up for connection to missed event. decision making Care Episodes Monitors variance from pre-defined episodes of care. process. •Monitors activities of predefined episodes of care to avoid overutilization of services and incurrence of unnecessary costs. Episodes are grouped by disease type (Coronary Heart Failure, Chronic Obstructive Pulmonay Disease, Heart Attack, etc.) Risk Scoring Attaches risk score to each patient based on severity of illness and presence of comorbidities. •Creates opportunity for providers to adequately distribute resources to patients most in need. High risk patients may depend on type of condition, medical history, demographic facts, compliance history, transition to home status, etc. This is a predictive modeling mechanism to help providers mitigate risk. Patient Registries Creates running database or list of patients by disease type to facilitate population health management •Allows providers to stratify patients to better understand the clinical dynamics of their disease and its impact on operations and finances. By grouping patients into groups such as high/low cost, high/low utilization, positive/negative outcomes, relationships between clinical activity and outcomes can be created to determine best practices as well as identify patients and processes that need attention. Continuity of Care Identifies when patients expose the system to risk by receiving Leakage care from provider's outside of system •For healthcare organizations that are focused on providing care for the entire patient continuum, when patients leave the system to receive care, the organization becomes exposed to risk. When patients receive care elsewhere, providers have no control over the types of care, outcomes or costs associated with that visit. By creating alerts, providers can be proactive in ensuring that patient outcomes are not jeopardized. By aggregating alerts, providers also gain insight into why patients are leaving the system (access, capacity, lack of follow-up, dissatisfaction, etc.) GIS Enabled Activity Creating instant "location effect" by mapping operational , Mapping market and competitive data •By placing operational data onto a GIS enabled map, healthcare organizations can instantly see how their activity interacts across its primary and secondary service areas. This provides the organization with insight on service area demand, capacity, performance, competitive advantage/disadvantage, demographic alignment and several other location-based.
- 29. CHALLENGES • Healthcare Organizations are Overwhelmed with IT priorities • EMR implementation, training and troubleshooting is a huge task • Data is aplenty and very much unalike • Structured and unstructured data will make integration difficult • Cultural Barriers will slow the buy-in and uptake process • Business units feel ownership of data, threatened by increased access • People are naturally resistance to change • Bringing “science” to decision making will take time for people to adopt • Reimbursement is not a certainty • Data may help with financial vitality but it is not the sole answer
- 30. FINAL THOUGHTS • Governance will play critical role in making BI a reality • All decisions are highly scrutinized and assessed from the highest levels of the organization • Data has revolutionized many industries • Healthcare is next on the innovation curve “in times of great change, it is the learners who inherit the future, the learned usually find themselves equipped to live in a world that no longer exists” - Eric Hoffer, Reflections on the Human Condition