210 likes | 359 Views
Squeezing Juice from Clinical Data Repositories: Information for Patient Management and ABF Revenue. Susan Smith Cardiothoracic Surgical Clinical Information Service The Prince Charles Hospital, Queensland Health, Brisbane Ian Smith St Andrews Medical Institute
E N D
Squeezing Juice from Clinical Data Repositories: Information for Patient Management andABF Revenue Susan Smith Cardiothoracic Surgical Clinical Information Service The Prince Charles Hospital, Queensland Health, Brisbane Ian Smith St Andrews Medical Institute St Andrews War Memorial Hospital, Uniting Healthcare, Brisbane
Background A variety of Clinical Information Systems (CIS) now exist • Operational & patient management systems eg • Medical imaging, diagnostics • Pathology • EMR – ED, Anaesthetics, Operating Theatre, Oncology, GP/Community, etc • Managerial/tactical • Bookings/referral systems • Strategic • Registries, vital statistics • Research databases
Background Increasing pressure for secondary use of clinical information to support decision-making due to a number of drivers: • Health System Reform, Restructure & Transformation • Information Revolution • ‘Evidence-Based’ paradigm • Accountability & Performance monitoring • Q&S • Multidisciplinary Research Activities • New Analytical tools
Background Increasing Development of analytical tools/technology eg • Data Integration • Data Warehousing • In-line memory • Hadoop • Business/Clinical Intelligence • Interactive reports • Dashboards • Analytics • Statistical Process Control for Healthcare • Geospatial Analytics • Visual Analytics • Data Mining • Predictive statistics
_________ Background For the process of secondary data use to support decision-making to occur we need to extract meaningful information from growing stores of data
Cardiothoracic Surgery Clinical Information Service Purpose of CSCIS • Provide accurate and reliable, clinically actionable information, from data available relating to Cardiothoracic surgical practice, to support/facilitate the best patient outcomes • Primary functions relate to: • Outcomes Data Acquisition, • CTSx Morbidity & Mortality Peer Review Reporting/Support, • Clinical Audit Reporting, • Supporting Clinical Research cohort definition, • Support retrospective observational analyses
Cardiothoracic Surgery Clinical Information Service To perform these duties CSCIS have: • Data Registry Database & Ancilliary data repositories, DLU • Tools – Access, Excel, SPSS, QI Macros • Clinical Informatics • Clinical Knowledge & experience (RN x3, Hosp scientist) PLUS • Health Informatics knowledge and experience (HIMOx2, MHlth Sci (CDM), M Epi) PLUS • Public Health/Epidemiology/Biostatistics skills (Outcomes /Audit/ analysis reporting, SPSS training)
Cardiothoracic Surgery Clinical Information Service “Succeeding with … analytics requires a database and information infrastructure that supports it, plus a culture that bridges the gap between DBAs and analysts” Wayne Eckerson, Director of Research for The Data Warehousing Institute - Assuming that the gap between analysts and clinicians is also bridged! - Socio-technical and cultural issue
Cardiothoracic Surgery Clinical Information Service Two examples of extending the use of registry-based information: • Activity Based Funding DRG coding Audit against Clinical Data • Analysis of Trends in Reoperation for Bleeding post CABG
1. ABF Project Data Audit and exchange with Medical Records FACT group to optimise accuracy of DRG allocation Levels of crosscheck for data capture • Cardiac Surgery Level- • Referencing against Clinical CTSx Register data, crosscheck DRG allocations to identify any inconsistent with cardiac surgery codes • eg cost difference: from $1,750 - $40,960 • Procedure Level - • confirm Valve, CABG, Other CTSx eg all concomitant procedures, complexity of procedure, Other CardThor related DRG appropriate • eg cost difference: from $1,750 - $40,960 • Complication Comorbidity Levels (CCLs) - • Sort Clinical Data records by Euroscore Risk Score (Clinical Severity index) • Cross check Clinical data with DRG coding for records with Euroscore >8% • If high risk score cases not coded to appropriate DRG codes, check for capture of comorbidities • Invasive investigations Level • eg sort Clinical data for inpatient preop coronary angiogram procedure, crosscheck against DRG allocation • eg cost difference: up to $13,456
1. ABF Project This could be done by audit of DRG output against charts, however this would be more costly and by use of the clinical data we can target the critical procedures, rather than review all cases. ie Pareto’s Principle or Juran’s observation of "vital few and trivial many": 80% of cost due to 20% errors
1. ABF Project Requirements: • Clinical Data repository + DRG expertise + Clinical DM expertise • Good relationship with Med Recs • Time allocation – approx 2 hrs per month (for 1350 cases (1200 discharges) pa of CTSx complexity) Limitations: • Nb some complications are inherent to particular DRGs ie can’t double capture • Still require Med Recs to correct coding Coding module • Time delay /can’t change after submission • Can’t audit everything Estimated Benefits: • Estimated increased revenue identified Jul-Jan: at least $200,000 • Audit feedback noticeably improves coding quality
2. Analysis of Trends in Reoperation for Bleeding post-CABG. Bleeding is a significant consequence of Cardiac Surgery. • Reported rates vary from 2-8% • TPCH identified increasing rate over 2002-2010 through regular peer review M&M meetings How do we use our data to investigate this? • Highly complex mechanism with many predictors and confounders • Physiological patient factors • Procedural factors • Care management factors
2. Analysis of Trends in Reoperation for Bleeding post-CABG. Data Issues: • Some good quality data • eg primary outcome: reoperation • Incomplete data on potential modifiers of bleeding rates eg • Preop antiplatelets therapy (poor documentation) • Use of antifibrinolytics – aprotinin, aminocaproic acid, TXA (not in CTSx Register – imprest stock) • Other system modifiers such as use of clinical pathway not captured (eg ACS)
2. Analysis of Trends in Reoperation for Bleeding post-CABG. Analysis methodology options • Traditional multivariate regression does not reveal factors that explain the increasing trend, difficult to discern trends for different procedures, etc • Statistical Process Control • Much used in industrial and engineering processes • Being adopted more widely in healthcare • Exponentially Weighted Moving Average (EWMA) statistically robust and clinically intuitively interpretable
2. Analysis of Trends in Reoperation for Bleeding post-CABG.
2. Analysis of Trends in Reoperation for Bleeding post-CABG.
2. Analysis of Trends in Reoperation for Bleeding post-CABG. Multivariate regression Odds Ratios for predictors of reoperation for bleeding following isolated CABG, 2002-2005. Expected risk (green) with observed (blue) reoperation for bleeding following isolated CABG, 2002-2005.
Non-Elective Elective
2. Analysis of Trends in Reoperation for Bleeding post-CABG. Requirements • Analytical tools: Excel, SPSS, QI Macros • Expertise: Clinical Data Management, Epidemiological, Statistical Process Control methodology • Resources: fte, financial & clinical support Limitations: • Indirect /circumstantial evidence • Caveats re data quality Benefit: • How can this enhance clinical decision-making? • How can this direct further work?
Conclusions Registry collected data can support a variety of uses Requires appropriate tools, expertise and resources Can be shown to have tangible and intangible benefits