The clinical data pipeline Fueling analytics for researchers and physicians from a single source

Transcription

1 The clinical data pipeline Fueling analytics for researchers and physicians from a single source

2 Contents Introduction 4 Enabling access to data 6 The data trust for clinical operations 8 The data trust for research 10 Roadmap for change 12 Conclusion 14

3 The reliable and efficient access to information from the Data Trust can enable health care organizations to achieve better performance in both their patient care and research missions. The fast adoption of Electronic Medical Records (EMRs) and other electronic patient information systems has dramatically increased the pool of data potentially available for analysis. At the same time, pressures to improve the cost and quality of health care, as well as a strong focus on improving the productivity of translational research, has generated demand for clinical data and analytical capabilities that has overwhelmed many IT departments in health care organizations. A health care organization s future achievement can depend on compliance with external measures such as Meaningful Use, PQRS, and Core Measures, not to mention the ability to drive clinical and operational performance to meet the demands of Pay-for-Performance (P4P), Accountable Care Organization (ACO), and other risk-based and value-driven payment structures. For institutions engaged in clinical research, the National Institutes of Health s (NIH) Clinical and Translational Science Awards (CTSA) program demonstrates that maintaining a competitive research program requires enabling investigators to have HIPAA and IRB-compliant access to patient care data as a research asset. The critical fuel required by both clinical operations and translational research is high quality clinical data. The Recombinant Data Trust is a demonstrated clinical data warehouse designed to centralize the management of patient information for analytical purposes. The resulting reliable and efficient access to data can help health care organizations to achieve better performance in their patient care and research missions, with higher quality care, improved operational performance, and productive and innovative translational research. The clinical data pipeline 3

4 Introduction The demands for data access and analysis in health care are rapidly growing in complexity and volume. Physicians and other providers, clinical operations management, and research investigators are all constrained by barriers to managing, accessing, and analyzing patient data that is generated and housed in Electronic Medical Records (EMRs) and other patient care systems. The architecture and data storage methods of these systems may not be well suited for using data beyond point of care transactions. The conventional approach to making clinical data available for secondary uses is to meet specific requirements by executing individual projects to extract data as needed from source systems. But this can result in overlapping projects with redundant efforts, inconsistent and poor quality data delivered to different projects, and heightened risks to compliance with HIPAA, IRB, and data security requirements. Accordingly, users are unable to have a high degree of confidence in the analytics available to them, and stakeholders are generally unsatisfied with the status quo. For secondary uses such as aggregate analysis for quality reporting, operational improvement, or translational research, data should be extracted from different patient care systems and managed in a central repository. But few health care organizations have repositories that systematically extract and integrate data from EMRs and other patient care systems to make it available for secondary uses. While building a true clinical data repository to support operational reporting or to support research can be a significant step forward for an institution, it may also be a major missed opportunity properly designed and executed, a single data repository can serve the needs of both the patient care and research missions of a health care organization. The conventional approach to data access is to execute individual projects for specific needs, which results in redundant efforts, inconsistent results, and risks to data security. Cardiac risk biomarkers study Research Epidemiology of pediatric obesity Infliximab RA clinical trial New investigational drug study Tissue bank Bio repository? New decision support framework Tumor registry Back surgery comparative effectiveness Genomic SNPs for breast cancer Pediatric Rhematology network Source data Outpatient EMR Inpatient EMR Billing Claims Labs Clinical trials Cardiovascular registry Balanced scorecard Diabetes registry Pay for performance/ contracting Quality PQRS Preventive health registry Safety/Harm management Provider incentive program Tobacco registry Accountable care organization Patient satisfaction Meaningful use Medication recalls 4

5 The Data Trust delivers cleansed and normalized data optimized for a variety of analytical purposes in both translational research and clinical operations Cohort selector Meta-registry manager i2b2 PQRS Selectrus performance analytics server Meaningful use The Data Trust centralizes data aggregation and cleansing for analytical applications to meet needs both in clinical operations and in translational research, leveraging data warehouse development and infrastructure investment and operational costs across both domains. The highest costs in analytics are generally the human resources to manage data interfaces and cleanse data, so maintaining separate efforts for clinical operations and research can be costly and wasteful. Because the Data Trust is specifically designed to deliver data for both clinical and research applications, this duplicate cost can be eliminated. C E R T I F I E D R E G I S T R I E S Sample trust Study enrollment Data subscription services Source adapters transmart Research Education Profiles RNS Data Trust Nucleus MDM Clinical Financial Data capture agents Speciality reporting modules Analytic engine Analytic datamart Report central HEDIS The Data Trust clinical data warehouse is a solution designed to meet the full spectrum of analytical and research needs by delivering cleansed and integrated data from EMRs, specialty systems for cardiology, radiology, GI, and other patient care areas, administrative systems (scheduling, claims, billing, etc.), and research systems. The Data Trust also incorporates a multi-tiered hierarchical security model, so operational and research applications can comply with HIPAA and HITECH requirements in the use of data. Data quality for analytics and research can also be improved as the data from different patient care systems is integrated through a standardized process of data extraction, cleansing, normalization, transformation, and loading (ETL) designed specifically to handle clinical data. The resulting consistency of data may also increase user confidence in analytic and reporting tools, as a single data repository can eliminate the conflicting results that are often reported by separate projects using different data sources or different ETL processes. The Data Trust is also designed to improve data security and regulatory compliance. Unlike other industries, where data security is most often a matter of protecting proprietary information for business purposes, data breaches in health care are subject to civil and criminal penalties under the provisions of HIPAA and HITECH. In the research domain, IRB requirements add another level of compliance risk. The traditional approach of individual data extracts feeding disparate project data stores can make consistent oversight and security extremely difficult. With a single resource serving data users across the enterprise, security and privacy compliance can be easier to maintain. The clinical data pipeline 5

6 Enabling access to data The adoption of EMRs and other electronic patient care systems has accelerated significantly in recent years as a result of Meaningful Use incentives and other forces in the health care environment. At the same time, the need for sophisticated analytics is increasing due to the growing number of external reporting mandates, more intense competitive pressures to improve and report operational performance, the development of risk-based and value-driven reimbursement under Accountable Care Organization (ACO) and Pay for Performance (P4P) programs, and implementation of care models like ACOs and the Patient Centered Medical Home (PCHR) that often cross traditional organizational boundaries. While data available to support these analytic needs is theoretically increasing with the growing use of EMRs and other systems, in practice this may not be the case launching and supporting EMRs and other new patient care systems has consumed the bulk of IT resources in many health care organizations, which prevents IT personnel from enabling access to the sizeable pool of new data being generated. Enabling access to patient care data for analysis and reporting requires addressing two specific challenges. First, the data architecture of EMRs and other transactional systems is not well suited to aggregate data analysis. Second, the complexity of clinical data requires specialized knowledge to cleanse and integrate data so meaningful analytics are possible. EMRs and other patient care systems are transactional systems created for high volumes of activity focused on individual patients at the point of care. These systems are designed to handle a complex set of data cataloging human health issues, providing a deep view of a single patient. The database schemas supporting these systems are managed for these individual patient-centric actions, and are not well suited to accessing data for aggregate analysis. In addition, directly reporting large volumes of data within transactional applications raises the risk of performance degradation and downtime in mission-critical patient care, which can be costly in terms of compromised patient safety and reduced productivity. The Data Trust can reduce this risk by managing data for analytics in a repository that is separate from transactional systems, with a design created for cross-patient queries. A single extract from patient care systems to the Data Trust can support multiple needs that otherwise may have required separate queries for individual projects. Even with frequent requests to reduce the latency of data available in the Data Trust, the overall query load on patient care systems can be reduced significantly. Once extracted from patient care systems, clinical data should be properly integrated to be analytically useful. Simply creating a replicated operational data store is not sufficient for health care analytics. Integrating clinical data is a complex process. Software must be developed to extract, cleanse, normalize, and transform data from a variety of source systems EMR, laboratory, pharmacy, scheduling, claims, billing, etc. Given the diversity of systems and the amount of customization that has usually occurred in their deployment, a deep understanding of these source systems and their data (coding standards, terminologies, and clinical context) is required to integrate data into a useful clinical data repository. An enterprise scale repository requires applying this understanding to developing software and audit processes to handle missing and inaccurate data consistently, and to address errors and discrepancies in data from different source systems. EMRs and other patient care systems are leveraged for individual patient-centric transactions, and are not well suited to accessing data for aggregate analysis. 6

7 Clinical data must be properly integrated to be analytically useful. Simply creating a replicated operational data store is not sufficient for health care analytics. For example, different clinical applications may each have separate coding systems for the same medication. The Data Trust is designed to represent the medication consistently with a single identification across data from multiple clinical applications and codes using medical terminologies designed to standardize content within and across organizations. These inconsistencies might not appear to be a critical problem, but resolving them correctly is essential for meaningful analytics. In the simplest case, duplicate data can cause errors in analytical programs that count records. In more complex scenarios, reporting may be inaccurate if quality analytics are programmed based on one coding system and miss data generated in a source system using another coding system. More seriously, patients may be put at risk when a patient is not notified of a medication recall or a provider s intervention report is inaccurate because information is fragmented without a mechanism to tie together data about an individual patient and create an overall view that spans different source systems. Clinical operations and research stakeholders can benefit from a shared investment in ETL and data cleansing and normalization processes for data in the Data Trust. Patient information Patient visits Labs Meds Billing Genomic data Data cleansing & normalization Data Trust Clinical operations applications Translational research applications The clinical data pipeline 7

8 The data trust for clinical operations The specific complexity of clinical data often renders conventional Business Intelligence tools, by themselves, inadequate for sophisticated analytics in a health care setting. Hospitals and provider groups require a more specialized approach, referred to as Clinical Intelligence. Detailed Clinical Intelligence tools are only now becoming available, as a result of the growing need for health care analytics and reporting. But many health care organizations lack the necessary infrastructure and resources to manage and distribute data to effectively deploy and use Clinical Intelligence applications. The Data Trust can fill this gap, providing a data pipeline to support the requirements of Clinical Intelligence applications for providers, operations, and management. The Data Trust is designed to integrate and manage the data generated by electronic patient care systems, and can fuel a clinical intelligence solution like the Recombinant Selectrus Performance Analytics Suite. The Selectrus Suite is designed specifically for clinical analytics and reporting, and can enable a health care organization to meet the demands of external reporting requirements, as well as managing and improving patient care. Recombinant s Selectrus Performance Analytics Suite The Selectrus Performance Analytics Suite is a framework for Clinical Intelligence, providing analytics and reporting for quality metrics, clinical performance improvement, and operations management. The Selectrus suite includes data models, analytical processing, and a library of reports and dashboard views designed specifically for the needs of health care organizations. The modular design of the Selectrus suite can provide for incremental implementation, and can enable conventional Business Intelligence tools to serve as a presentation layer for Selectrus. The data available from the Data Trust can enable analysis and reporting for Meaningful Use, PQRS, Core Measures, JCAHO, and the growing number of other metrics that health care organizations must report to government agencies, accrediting bodies, and payers. Although modules for EMRs and other source systems are available to support some of these requirements, an approach based on the Data Trust supports reporting and analysis spanning the entire organization. This is especially valuable when phased implementation of new systems or the desire to retain a legacy system would otherwise result in fragmented reporting from different systems covering various parts of the organization. Recombinant s Meaningful Use Registry, for example, is designed to use data from the Data Trust (or other clinical data warehouse) to track Meaningful Use measures, and to calculate and submit clinical quality measures to CMS. Since the Data Trust can integrate data from multiple EMRs and other patient care systems, the Meaningful Use Registry provides consolidated Meaningful Use results for institutions that have different systems deployed for different parts of the organization, or institutions that want a more flexible Meaningful Use strategy independent of a specific EMR vendor. Data integrated from multiple source systems can be crucial to the effectiveness of chronic disease management, preventive care, and other population health initiatives. With data from different systems and different providers that offer overlapping views of an individual patient, the Data Trust can enable analytical tools to use data from one source to cover the gaps in another source. In the past, initiatives like chronic disease management have relied primarily on payer billing data, which has been shown to be unreliable for clinical purposes. The Data Trust is designed to integrate clinical data with billing data, and along with the Master Patient Index and Provider Master metadata managed by the Data Trust, can enable a more detailed view of an individual patient, and more deliberate attribution of patients associated with specific providers. These can enable better patient management and program reporting to help improve the results of population health 8

9 initiatives. Recombinant s Selectrus Performance Analytics Suite incorporates clinical measures logic to calculate metrics such as compliance with National Quality Forum (NQF) metrics for diabetes and other chronic diseases allow organizations and individual providers to measure their performance against accepted standards of care. The Selectrus Analytics Engine also facilitates development of custom metrics to support institution-specific initiatives that can evolve over time. ACOs, P4P, and other approaches to risk-sharing and value-based reimbursement require data sourced from across functional and departmental boundaries within organizations, and in the case of ACO, across organizational boundaries that separate different provider groups. The Data Trust is designed for exactly this type of data integration, and its security model can also enable data protections to address competitive and other concerns of provider organizations participating in an ACO or other arrangements requiring data sharing and collaboration. The Data Trust security model, and its provider and organizational metadata enable the Selectrus Suite to present roll-up reports that allow administrators to assess the performance of their operational units, and to drill down to specific provider and patient-level data to identify specific performance issues and potential risks. Similarly, the Selectrus Suite reports allow providers to benchmark their performance relative to the rest of the organization, and to drill down to identify specific actions required for individual patients. The clinical data pipeline 9

10 The data trust for research Access to patient care data has long been a difficult challenge for researchers. Individual patient data is protected by HIPAA and IRB standards and procedures. Researchers typically face a time-consuming process for data access requiring development and approval of an IRB protocol, and they must often work through overtaxed biostatisticians or IT analysts to actually obtain data. In many institutions, data retrieval is done with inefficient manual processes, based on individual reviews of patient records and correspondence with providers. When electronic resources are available, multiple custom queries must be run directly against source systems to gather the information needed. The redundancy of these multiple queries for different investigators can create an unnecessary burden for analysts retrieving data, and inconsistencies across different systems can create significant challenges to integrating data. Caution about overloading mission-critical patient care systems with data retrieval queries can also delay the availability of data for researchers. In the end, data for researchers is often not delivered in a timely manner, and when provided it is often incomplete, inaccurate, and/or reported in a manner that limits its value for individual and aggregate research purposes. As a result, investigators are unable to efficiently take advantage of the tremendous potential of patient care data as a resource for research. It is widely recognized that the current ad hoc approach to accessing patient data has had a negative impact on biomedical research. The Clinical and Translational Science Awards (CTSA) program of the National Institutes of Health is a nation-wide initiative aimed at accelerating the translation of scientific discoveries to patient care and improved health outcomes. One of the objectives of the CTSA program is to help Academic Medical Centers (AMCs) build functional data repositories of clinical data to increase the speed of innovation and productivity of clinical research. A clinical data repository accessible to researchers has effectively become a requirement for a competitive biomedical research program. The Data Trust can offer multiple levels of utility for translational research in meeting this requirement. It can accelerate cohort discovery and hypothesis generation, enhance researcher productivity and creativity, improve The Data Trust is designed to address the data access needs of both research and clinical operations, so it offers an incentive for clinical stakeholders to participate in a shared data repository project. the quality of research data, support population-based research projects, and strengthen compliance with HIPAA and IRB requirements. Because the Data Trust is specifically designed to serve the data access needs of both research and clinical operations, it also offers an incentive for clinical stakeholders to participate in a shared data repository project that enables researchers to have access to data from clinical patient care systems. This can be a valuable bridge that improves overall collaboration between the clinical and research missions of a health care organization. The i2b2 translational research application can enable organizations to maintain HIPAA and IRB compliance while allowing investigators to use data from patient care sources. The leading practice implementation of i2b2 is to use the Data Trust clinical data warehouse to manage data extracted from patient care systems, and to load a de-identified or limited data set into an i2b2 datamart from the Data Trust in accordance with the organization s patient privacy and research consent policies. Researchers can conduct more effective preparation-to-research type activities on the fly, with rapid cycles of query and analysis of the i2b2 dataset to discover patient cohorts and generate research hypotheses before finalizing grant proposals or seeking IRB approval. After IRB approval, data can be delivered from the Data Trust under the oversight of an Honest Broker process to facilitate compliance with patient privacy and research permission requirements. As a source of clinical data that has been cleansed, integrated, and organized specifically for aggregate analysis, the Data Trust can provide researchers more consistent, detailed data than is typically obtained from ad hoc data extracts from different systems in response to project-specific requests. 10

11 Resources are recovered for high value analysis tasks by reducing time spent gathering and cleansing data. The individual project approach The Data Trust approach Lower value Data gathering Centralization simplifies collection Too much time spent gathering data Data gathering Data cleansing Analysis Data cleansing Single cleansing prevents duplicate efforts Inefficient redundant cleansing Wasted resources Analysis time gained Resource allocation by activity Analysis More time for through analysis and reporting Higher value Not enough analysis Because of the consistent data exploration and delivery workflow enabled by the Data Trust, researchers can focus on planning and executing research rather than searching and negotiating for data. Biostatisticians, data analysts, and IT Personnel are also able to devote more time to analytical work and other high-value activities, rather than retrieving data for researchers. And overall compliance with HIPAA and IRB requirements is improved with better oversight of data access and consistent application of appropriate privacy measures. The Data Trust can address clinical and research needs for data with a single investment, and is designed to provide a tested pathway to managing clinical data in the health care enterprise. The Data Trust can also aid researchers in executing clinical studies. The integrated data archived in the Data Trust is considered valuable in and of itself for research. Investigators can use data delivered from the Data Trust to study many different topics, including chronic diseases, provider adoption of new interventions or treatment guidelines, outcomes studies, and comparative effectiveness research. With IRB approval, identified data for patient cohorts discovered with i2b2 can be used for clinical study recruitment, accelerating subject enrollment and study completion. Electronic Data Capture (EDC) forms can be pre-populated with demographics and other patient-specific data from the Data Trust, improving the quality of research data and the productivity of study coordinators who currently enter data manually into Case Report Forms. Integrating research data into the Data Trust can make clinical studies an asset available for other purposes in the organization. Clinical study data can be loaded to a protected segment of the Data Trust, and made available to other projects with appropriate compliance oversight and stakeholder permissions. For example, specialized datamarts can be created that link research study data to the overall clinical view of patients enabled by the Data Trust, providing investigators with additional insights and discoveries. Or data from multiple projects conducted at an institution can be integrated in a new datamart that increases the ability of researchers to assess previous research in planning new studies, or to conduct new metaanalyses, which could also incorporate data obtained in the course of patient care. The value of biorepositories can also be leveraged with links to the Data Trust. Patient data in the Data Trust can be used for automating specimen annotations in biorepository systems, and specimen data can be loaded into the Data Trust and used to allow researchers to know about the potential availability of specific types of samples. One example use case is to include biospecimen data in an i2b2 implementation, so that tissue sample characteristics can be another query parameter in cohort discovery. The clinical data pipeline 11

12 Roadmap for change The Data Trust can be used to help solve problems in clinical data integration and management, and improve the quality and accessibility of data for secondary uses. Clinical operations and translational research require access to similar data, but their analytical and reporting applications are very specialized. A consolidated solution to make data more accessible for both clinical analysis and research purposes is needed at both small and large institutions. But an undertaking of this nature may appear daunting, especially to organizations that are already faced with pressing demands to deploy or upgrade applications for patient care and research. The path to better data management, accessibility, and quality is an incremental progression. Many enterprise data management projects have failed because their scope has been too broad and deliverables too distant to achieve initial value to stakeholders, leading to abandonment as major costs are incurred without corresponding benefits. Individual projects focused on delivering data for specific needs are more likely to achieve the intended benefits, but are limited in scope and often result in isolated data silos. To avoid these problems, institutions should build a clinical data warehouse one stage at a time. The Data Trust is specifically designed for such an incremental deployment strategy. This pragmatic approach begins with creating an overall roadmap that allows the institution to identify specific priorities at the enterprise level, and proceeds with executing focused, manageable projects to address those priorities with specific high value applications that have a high probability of success. This roadmap-based approach differs from creating ad-hoc departmental systems because the roadmap provides a plan for integration as the final goal. An incremental approach to building the Data Trust avoids the risks of both ad-hoc and big-bang development strategies. Establish overall roadmap Identify and execute targeted projects to address key priorities Integrate projects into an enterprise architecture 12

13 Seemingly disparate projects may quickly deliver value to their champions, but like pieces of a jigsaw puzzle, they will likely fall into place and coalesce into an enterprise data architecture. The cleansed data and reusable ETL processes resulting from one project become shared components that can accelerate the execution of additional projects. This piece-by-piece expansion of the Data Trust with new data sources and components to support additional analytical needs can build sustained momentum for the enterprise strategy. Periodic reviews of the roadmap can allow organizational leadership to respond to changing circumstances by revising priorities and updating project objectives. The iterative process of review, reprioritization, and execution helps to make certain the roadmap remains synchronized with the organization s strategy, and that resources are regularly reallocated to address the organization s most important needs as circumstances change over time. The clinical data pipeline 13

14 Conclusion Successfully utilizing clinical data is critical to the ability of health care organizations to deliver high-quality care and maintain productive research programs. Health care organizations should address the growing demand for data to fuel analytics for clinical operations and translational research. The conventional approach to this problem has been to execute individual projects by extracting data as needed from source patient care systems. But this can result in redundant efforts, performance degradation of mission-critical patient care systems, inconsistent results with unsatisfactory data quality for users, and risks to compliance with HIPAA, IRB, and data security requirements. On the other hand, large scale enterprise data warehouse projects often fail because of the inability to build and generate momentum with short-term deliverables that engage stakeholders. The Data Trust is designed to address these issues, enabling an organization to meet clinical and research needs for data with a single investment. The detailed design of the Data Trust can allow smaller, focused projects to proceed with an overall plan for integration into the overall enterprise architecture. The Data Trust is also designed to be implemented with a pragmatic, incremental approach, giving organizations the ability to avoid the risks of big-bang enterprise scale projects. With the Data Trust as a detailed source of cleansed and integrated data, health care organizations can be positioned to better respond to external reporting mandates, manage participation in new reimbursement structures, and maintain strong a strong competitive position with continuous improvement in clinical and operational performance. With effective research access to data through the Data Trust, organizations can improve the opportunity for investigators to conduct innovative clinical studies while improving data security and compliance with HIPAA and IRB requirements. Utilizing clinical data is a critical factor in the ability of health care organizations to deliver high-quality, cost effective care, and to maintain productive, innovative translational research programs. The Data Trust can provide a demonstrated pathway to achievement in managing clinical data in the health care enterprise. 14

15 The clinical data pipeline 15

16 This publication contains general information only and is based on the experiences and research of Deloitte practitioners. Deloitte is not, by means of this publication, rendering business, financial, investment, or other professional advice or services. This publication is not a substitute for such professional advice or services, nor should it be used as a basis for any decision or action that may affect your business. Before making any decision or taking any action that may affect your business, you should consult a qualified professional advisor. Deloitte, its affiliates, and related entities shall not be responsible for any loss sustained by any person who relies on this publication. About Recombinant by Deloitte Recombinant delivers innovative data warehouse and clinical intelligence solutions to academic medical centers, health care provider organizations, biopharmaceutical companies, and other users of clinical and research data. Recombinant is focused entirely on secondary uses of health care and life sciences data, providing expert consulting in data strategy and governance, software products for clinical data warehousing, analytics, and reporting, and professional open source services for implementation and support of open source informatics applications. To learn more about Recombinant s technology and services for health care data warehousing and clinical intelligence, please contact us at or results@recomdata.com. For more information about Recombinant by Deloitte, visit About Deloitte Deloitte refers to one or more of Deloitte Touche Tohmatsu Limited, a UK private company limited by guarantee, and its network of member firms, each of which is a legally separate and independent entity. Please see for a detailed description of the legal structure of Deloitte Touche Tohmatsu Limited and its member firms. Please see for a detailed description of the legal structure of Deloitte LLP and its subsidiaries. Certain services may not be available to attest clients under the rules and regulations of public accounting. Copyright 2012 Deloitte Development LLC. All rights reserved. Member of Deloitte Touche Tohmatsu Limited