CaisisPresentation20 06

The Caisis System: 

The Caisis System Paul Fearn Urology/GU Informatics Manager Department of Surgery, Urology Service Memorial Sloan-Kettering Cancer Center www.caisis.org November 8, 2006

The MSKCC Caisis Team: 

The MSKCC Caisis Team Paul Alli – Database Administrator Avinash Chan - Developer Jason Fajardo – Lead Designer / Developer Paul Fearn – Project Leader Kevin Regan – Project Manager / Developer Beth Roby – Administrative Frank Sculli – Lead Developer Brandon Smith – Developer Jason Stasi – Research Project Coordinator Kinjal Vora – Research Data Coordinator

What is Caisis?: 

What is Caisis? Web-based database application Integration of clinical and research activities Chronological list - “Date/Variable/Value” Data processing algorithms Accuracy Reproducibility Standardized and automated workflows Open source license and development

“Date-Variable-Value” Concept: 

“Date-Variable-Value” Concept Chronological list Partial or “fuzzy” dates Data Source/Quality Audit Log “Lock” records Calculate key variables: (baseline, progression, outcomes)

Dataset Production Algorithms: 

Dataset Production Algorithms Automated nomograms and progression calculations

Why did we start Caisis effort?: 

Why did we start Caisis effort? Improve data quality Reduce duplication of effort Increase research productivity Reduce RSA skill specificity Overcome scale limits of Access databases and spreadsheets Promote innovation and collaboration

What is high quality data?: 

What is high quality data? Accurate, and backed up by source documents Up-to-date Quick and easy to find Free from duplication, (for example, where two or more different records exist for the same patient) Free from fragmentation, (where different parts of a patient's records are held in different places, possibly in different formats

Design Issues for Scalability: 

Design Issues for Scalability All databases grow in size and scope “Spreadsheet” may be more costly in long run Limits to process improvements Staff entrenchment or turnover problems Risks of data corruption and data loss

Business Problems: 

Business Problems Research costly - staff time and effort Single institution results not reproducible Need standard or interoperable data models Need transparent data processing algorithms Cannot do next generation research without inter-institutional collaboration Need large, clean, minimally biased datasets Need open source code for innovation

Systemic Problems: 

Systemic Problems Frustration with user interfaces of existing databases Proliferating fields (i.e. giant spreadsheet syndrome) Investigator biases built into systems Specialized knowledge and experience required (entrenched data managers) to prepare datasets for research Staff size required to maintain data not feasible when number of cases increases Too many screens required for data entry; no way to see patient’s integrated history Research CRFs are extra work for clinicians

Vision: 

Vision To bridge the gap between Clinical Research and Clinical Practice Collect research data at the point of service Electronic OR documentation Electronic Clinical documentation Generate compliant documentation Automate or verify billing levels Automate patient disposition Improved data quality and quantity to further research goals

Plan: 

Plan Develop a robust, scalable database: Caisis Incorporate data collected for research into the clinical workflow Short Term: Populated paper forms Long Term: Electronic forms (eForms) Standardize method for collecting data Integrate process with hospital medical records and billing

Requirements: 

Requirements Impact on clinical flow must be minimal Quality of patient care must be maintained System must be easy to use yet flexible enough to document clinical events Integrate clinical decision making tools: Nomograms, Graphs, PSA Velocity, and Doubling Time Support research objectives

The database is not the goal!: 

The database is not the goal! Fundamental Theorem of Biomedical Informatics Friedman CP, Wyatt JC, Evaluation Methods in Biomedical Informatics, 2nd ed + >

Physician Acceptance of IT: 

Physician Acceptance of IT “To be widely accepted by practicing clinicians, computerized support systems for decision making must be integrated into the clinical workflow. They must present the right information, in the right format, at the right time, without requiring special effort. In other words, they cannot reduce clinical productivity.” – Brent James, NEJM 2001

Slide16: 

A chronological patient history composed of common data elements (CDEs) for clinical practice and research Common Data Elements

Common Data Elements: 

Common Data Elements Cumulative Factors: Social History, Family History, Race/Ethnic Group

Prostate Cancer Common Data Elements: 

Prostate Cancer Common Data Elements Patient’s demographics: MRN, DOB, Ethnicity, Family Hx. All relevant biochemical markers All physical exam (DRE) results AJCC/UICC Clinical Tumor Stage by physician All relevant biopsies with pathologic findings All relevant diagnostic imaging studies and findings All relevant radiation therapy All relevant surgical procedures with pathologic findings All relevant medical therapy and concomitant meds All quality of life assessments (urinary/sexual/bowel function) Vital status by date

Algorithms…: 

Algorithms…

Longitudinal Follow-Up: 

Longitudinal Follow-Up SSDI batch queries Automation tools

Next Step: Populated Clinical Templates: 

Next Step: Populated Clinical Templates Collect structured data for research Billing compliant Better documentation of service Reuse standardized sections Populate clinic forms HPI algorithms to summarize clinical data Data feeds from institutional systems: labs, demographics, appointments, procedures

Web-based database application: 

Server Web-based database application Database Intranet or Internet Facility Firewalls Web Server Operating System App Server

“Valhalla” Version 1.0 - 1.1: 

“Valhalla” Version 1.0 - 1.1

“Valhalla” Version 1.2: 

“Valhalla” Version 1.2

Populated Clinic Forms: 

Populated Clinic Forms

Effects of Integration on Data Quality: 

Effects of Integration on Data Quality Populate clinic forms with data from research database Collect research data during clinical workflows Multiple people enter, review and correct data Identify outliers from research Fill gaps and correct errors Multiple stakeholders analyze, and correct data

Increasing Scale and ScopeIntra-Institutional Integration: 

Functional Areas Researchers Clinicians Nurses Fellows Administration Informatics Increasing Scale and Scope Intra-Institutional Integration Departments Urology Medicine Pathology Radiology Surgery Radiation Oncology Information Systems Computational Biology Diseases Prostate Bladder Kidney Testis Pancreas Colorectal Gastric Thyroid

HIPAA Privacy and Security: 

HIPAA Privacy and Security Confidentiality Ensuring that patient/institution confidentiality agreements are not compromised by operational and research activities Privacy Data which can identify a patient (HIPAA Identifiers) belong to the patient, and are only exposed to and used by authorized staff (roles) for authorized purposes Data Integrity Data must be protected to persist as the last authorized user left them System Integrity System protected against unauthorized use, accidents, natural disasters and malicious attacks

System Privacy and Security: 

System Privacy and Security Limited access to patient data by job function (role) and dataset. HIPAA compliant data export IRB approval or de-identification required Disclosures logged Tracking / Logging Who views which patient Who performs what action Nothing is overwritten (FDA full audit trail)

Development Platform Options: 

Development Platform Options Operating system Windows, Linux Database SQL Server, Oracle, DB2, MySQL, Access Web server IIS, Apache Web application platform ASP.NET, J2EE/JSP, ColdFusion, CGI Programming language C#, Java, PHP, Perl

New Application Platform: 

New Application Platform Microsoft ASP.NET Windows server operating system Microsoft IIS web server Microsoft .NET Framework Microsoft SQL Server Open source license (GPL) Free to use Free to modify source code Distributed modifications must open source

Caisis 2.0: 

Caisis 2.0 Integrated prostate, bladder, kidney, testis cancer Integrated Urology and GU Medicine Completed full suite of billing/compliant forms Migrated from ColdFusion to Microsoft.NET Re-architected for security, stability and scalability

Caisis Version 2.0 – 2.1: 

Caisis Version 2.0 – 2.1

Caisis 3.0: 

Caisis 3.0 Module framework Plugin framework Over 50,000 patients Protocol manager Image upload plug-in First eForm - Prostatectomy

Protocol Manager: 

Protocol Manager

Caisis 3.0 User Interface: 

Caisis 3.0 User Interface

Surgery and Clinic eForms: 

Surgery and Clinic eForms

eForms / Tablet PCs: 

eForms / Tablet PCs

Caisis Project Timeline: 

Caisis Project Timeline Microsoft Access databases 1999 ProstateDB 1.0 2000 PRDB / Prostabase ColdFusion & SQL Server web-based database 2002 Valhalla 1.0 – 1.1 Prostate 2003 Valhalla 1.2 (7,994 patients) Billing/EMR compliant populated clinic forms Microsoft.NET & SQL Server web-based database 2004 Caisis 2.0 – 2.1 (26,470 patients) Integrated bladder, kidney, testis 2005 Caisis 3.0 – 3.1 (44,000 patients) Prostatectomy eForm, protocol manager, tumor maps 2006 Caisis 3.5 – (> 55,000 patients) GU and Urology Prostate Follow-up eForms

Slide40: 

Lab QOL Survey Cystoscopy Biopsy Outside Provider MSKCC Surgery Pathology Pharmacy Radiology Lab Clinic Visit Cystoscopy Biopsy Surgery Pathology Pharmacy Radiology Patient Long F/U SSDI Government Clinic Form / Dictation CAISIS Report or Dictation EMR+OMS Registration Appointmnts IDB QOL DB

Inter-Institutional Collaboration: 

Inter-Institutional Collaboration MSKCC Seattle Consortium (Fred Hutchinson / Univ of Washington) Baylor College of Medicine University of Rochester Wayne State University (Karmanos Cancer Institute) Westmead (Sydney) Cancer Research UK (London) University of Malmö Case Western Reserve University Cleveland Clinic University of Texas – San Antonio University of Texas Southwest Medical Center New York Prostate Institute

2005 Caisis Usage Statistics: 

2005 Caisis Usage Statistics

Supplemental FundingBISTI R01 Grant (2006-2009): 

Supplemental Funding BISTI R01 Grant (2006-2009) Extensibility Restructure procedures tables (and others) Implement “Meta-data driven fields” caBIG Silver Level compatibility Integration of Microsoft Access prototyped components Database browser, longitudinal f/u, project tracking Algorithms, nomograms Ease of Use Documentation and web site improvements Enhancements to administration / configuration tools Portability Macintosh web browser compatibility

BISTI R01: Year 1 Overview: 

BISTI R01: Year 1 Overview Restructure the data model to accommodate other diseases Remodel procedures, pathology, physicians Meta-data driven fields (entity-attribute-value) Dynamic or hybrid data entry forms Migrate Microsoft Access prototype from CaisisDB.mdb Database browser tool Dataset production algorithms - prostate Longitudinal patient follow-up workflows Prepare Caisis for caBIG Silver level compatibility Data Elements to caDSR Macintosh web browser compatibility

Return on Investment: 

Return on Investment The database itself is not the goal! Focus Caisis development and operations to achieve clinical practice and research goals Automate costly manual processes Improve staff resource flexibility Correct problems upstream to reduce waste of time and effort Develop integrated data supply chain

Costs of using Caisis: 

Costs of using Caisis Hosting the application and database Data entry / database management Creating queries and reports Customization / programming

Detection and Correction CostsData quality management: 

Detection and Correction Costs Data quality management Like cockroaches or rats that you actually see in the NY subway, visible data errors may be symptoms of widespread problems Regular audits for accuracy compared to medical record Thorough investigation of problems Follow-up measures to correct problems Follow-up measures to prevent future problems Coordination of activities Weekly Caisis operations meeting Ongoing training and management of staff

Acknowledgements: 

Acknowledgements Paul Alli Bernard Bochner, MD Ophelia Chiu Jason Fajardo Tia Higano, MD Michael W. Kattan, PhD Jared Katz David Kuo Hans Lilja, MD Louis Potters, MD Kevin Regan Victor Reuter, MD Beth Roby Peter T. Scardino, MD Howard I. Scher, MD Frank Sculli Hee Song Seo Mark Snyder Jason Stasi Kinjal Vora

Questions and Answers: 

Questions and Answers