DATA QUALITY MANAGEMENT THE FUNDAMENTALS FOR TODAY’S SUCCESSFUL HEALTHCARE ENTERPRISE WVHIMSS FALL CONFERENCE NOVEMBER 14, 2014 AGENDA • DATA 101 • DATA 911 • DATA QUALITY EXPLAINED • UAB HEALTH SYSTEM – A CASE STUDY • LESSONS LEARNED AND CALL TO ACTION • DISCUSSION DATA 101 • HOW MUCH NEW DATA DOES YOUR ORGANIZATION CREATE IN A DAY? FROM WHICH SYSTEMS? • WHERE DOES YOUR BUSINESS-CRITICAL DATA RESIDE ACROSS THE ENTERPRISE? WITH WHOM IS IT SHARED? • DO YOU COLLECT AND STORE THE RIGHT INFORMATION TO SUPPORT THE ORGANIZATION? • IS THE RIGHT DATA AVAILABLE TO SUPPORT TIMELY DECISION-MAKING? • IS THE DATA CONSISTENT WITHIN AND ACROSS THE ENTERPRISE? • CAN YOU GENERATE PERFORMANCE IMPROVEMENTS THROUGH DATA INSIGHTS? DATA 911: WHY HEALTHCARE DATA IS ESPECIALLY COMPLEX AND DIFFICULT TO MANAGE • LOCATION: • DEFINITIONS: • FORMAT: • COMPLEXITY: • STRUCTURE: • REGULATORY REQUIREMENTS: DATA 911: COMMON CULPRITS OF BAD DATA QUALITY • ARCHITECTURE AND APPLICATION COMPLEXITY: • LACK OF OWNERSHIP AND RESPONSIBILITY FOR DATA QUALITY: • REPETITIVE OR AMBIGUOUS BUSINESS PROCESSES: • UNCLEAR AND MULTIPLE DEFINITIONS OF DATA ELEMENTS: • NO CLEARLY DEFINED DATA QUALITY ESCALATION PROCESS: WHY DOES DATA QUALITY MATTER? GOOD DATA IS KEY TO: • EFFECTIVE DECISION-MAKING AT TO THE BOARDROOM EVERY LEVEL: FROM THE BEDSIDE • STRATEGIC PLANNING • CLINICAL QUALITY • HEALTHCARE OPERATIONS AND FINANCIAL MANAGEMENT • PUBLIC POLICY • PUBLIC AWARENESS ABOUT FACTORS THAT AFFECT HEALTH DATA QUALITY FRAMEWORK DATA QUALITY IS ONE Assess COMPONENT OF A DATA MANAGEMENT Remediate PROGRAM Monitor •Discovery •Profiling •Analysis •Cleansing •Enhancement •Consolidation •Measure and Compare ASSESSING DATA QUALITY • DATA QUALITY IS A MULTI-DIMENSIONAL CONCEPT • MEASURE. USABLE METRICS ARE REQUIRED TO DEFINE AND ASSESS QUALITY. • DATA CAN BE ASSESSED BOTH OBJECTIVELY AND SUBJECTIVELY • CONTEXT IS OFTEN IMPORTANT WHEN ASSESSING QUALITY • QUALITY IS DEFINED BY THE INSTITUTION AND DOES NOT NECESSARILY MEAN PERFECTION DIMENSIONS OF DATA QUALITY QUANTITATIVE DIMENSIONS QUALITATIVE DIMENSIONS OTHERS COMMON DIMENSIONS • ACCURACY • COMPREHENSION • ACCESS • CONSISTENCY • RELEVANCE • UNIQUENESS • TRUST • LINEAGE • OBJECTIVITY • VOLUME • SECURITY • INTERPRETABILITY • REPUTATION • COMPLETENESS* • EASE OF MANIPULATION • TIMELINESS* • MAINTAINABILITY • RATE OF DECAY UAB HEALTH SYSTEM A CASE STUDY IN DATA QUALITY PROVIDER DATA QUALITY PROVIDING MULTIPLE CHALLENGES TO THE ORGANIZATION. • DUPLICATE DATA • MISSING DATA • INACCURATE DATA UAB PROFILE • LARGE ACADEMIC HEALTH SYSTEM IN CENTRAL ALABAMA. • HEALTH SYSTEM COMPRISED OF MULTIPLE LEGAL AND OPERATIONAL ENTITIES • TWO PRIMARY IT ORGANIZATIONS + DEPARTMENTAL • HEALTH SYSTEM INFORMATION SERVICES (HSIS) • MANAGEMENT SERVICES ORGANIZATION (MSO) WHAT IS “PROVIDER” DATA? • • • • • • • DISCOVERY WHERE IS THE DATA? • • • • • • • • DISCOVERY WHERE DOES THE DATA COME FROM? WHERE DOES IT GO? • • DISCOVERY • WHAT IS DATA PROFILING? • PROFILING • DATA PROFILING DATASET STATISTICS • • • • PROFILING A DATA ELEMENT EXAMPLE DATA PROFILE: WHY DO WE CARE? • • WHAT WOULD WE EXPECT TO SEE? • DATA PROFILING EXAMPLE (COLUMN) NATIONAL PROVIDER IDENTIFIER Data Type: STRING The NPI is a numerical value. Identified as a String? Basic Counts Should there be Nulls? Type Count Null Non-Null Duplicates Distinct Non-Unique Unique Why are there duplicates? COUNTS Null, 19.31% % 4,576 19.31% 19,118 80.69% 69 .29% 19,049 80.40% 69 .29% 18,980 80.10% Duplicates , 0.29% Distinct, 80.40% DATA PROFILING EXAMPLE NATIONAL PROVIDER IDENTIFIER Pattern Frequency Value NULL Additional Counts Count % Type Value Frequency 4,576 19.31 Min Value 1000004138 1 DDDDDDDDDD 17,117 72.24 Median Value 1558551127 1 LDDDDDDDDD 2,000 8.44 Max Value T000199999 1 1 0.00 DDDDDDDDD NPI is a 10 digit identifier. Inconsistent findings Type Value Min Length 9 Median Length 10 Average Length 10 Maximum Length 10 DATA PROFILING EXAMPLE NATIONAL PROVIDER IDENTIFIER WHAT DID WE FIND? • • • DATA PROFILING EXAMPLE COMMON PROFILING ANALYSIS Average Value Mean Value Median Value Range Analysis Sparseness Cardinality Uniqueness Value Distribution Value Absence Minimum Value Maximum Value Standard Deviation Type Determination Format Identification Overloading Semantic Variance Analysis Key Analysis Dependency Analysis Orphaned Records Redundancy Analysis Integrity Analysis WHAT WAS LEARNED? • • • • ANALYSIS WHAT WAS LEARNED? • • • • • • • ANALYSIS WHAT WAS LEARNED? • • • • • • ANALYSIS ONGOING REMEDIATION • IT STEWARDSHIP INITIATIVES: • ACQUISITION OF NEXTGATE MDM SOLUTION • DATA CLEANUP • DE-DUPLICATION • NORMALIZATION • DEPARTMENTAL INITIATIVES (IT LED) • WORKFLOW OPTIMIZATION • INSTITUTE DATA GOVERNANCE (IT INFLUENCED) • DEFINE OWNERSHIP • DEFINE STEWARDSHIP • DEFINE ENTERPRISE SEMANTICS • SET GOALS, MEASURE, REVIEW, AND REACT DATA MANAGEMENT IS AN ORGANIZATIONAL ISSUE BEST PRACTICES IN DATA QUALITY MANAGEMENT • • BUSINESS OBJECTIVE DATA “TOXICITY” • TAXONOMY/DATA DICTIONARY • CLEANSE AND ENRICH DATA • SOFTWARE PLATFORM OR REPOSITORY • GOVERNANCE AND STEWARDSHIP HOW TO DELIVER DATA QUALITY • CREATE THE APPROPRIATE • GOVERNANCE RISK-BASED APPROACH – FOCUS ON THE AREAS THAT POSE THE GREATEST RISK TO THE BUSINESS • MAINTAIN BUSINESS BENEFIT – PERFECT DATA IS NOT THE AIM, ONLY DATA FIT FOR ITS PURPOSE • PROACTIVE – ADDRESS KNOWN ISSUES THAT DERIVE FROM POOR DQ, INVESTIGATE AND ASSESS IMPACT ON BUSINESS STRATEGY • ITERATIVE • INTEGRATED • SUSTAINABLE • FLEXIBLE WHERE TO START • • • • • QUESTIONS? Dan Rounds President Immersive darounds@immersivellc.com 888.869.0984 Immersivellc.com 3411 High Cliff Road, Panama City, FL 32409 Brian Bishop Operations and Interface Manager UAB Health System Information Services bebishop@uabmc.edu