bannerprint

In This Issue: Data Mining

This issue of the Clinical Informatics Insights focuses on Data Mining—with features covering information, access, privacy and security, acheiving quality outcomes and results. Be sure to check out the special feature, ONC Announces SHARP Awards: A Look at Secondary Use of EHR Data Research, and this month’s Tool Box Picks and featured discussion.

April 2010

State of the Industry

Privacy and Security Concerns in Data Mining

By Erik Pupo

Human medical data is seen to be one of the more rewarding and yet most difficult of all data to analyze. As providers grapple with an increasing amount of information from EHRs and other health IT data sources, data mining, with its promise to efficiently discover valuable information about patients from large amounts of information, is increasingly noticed as a technological breakthrough in the ongoing advance of health IT. Yet many privacy and security experts see data mining as the most fundamental challenge that consumers will face in the next decade. The specific challenge in data mining is developing precise models for analysis of data without providing access to the precise information in specific patient records. For example, prescription reminders sent from a drug store require access to specific information on a patient that may have been captured by a clinician in a database, but which the patient never consented to deliver to the drug store for data mining. In this specific example, does the promise of medication reminders outweigh the potential misuse of patient data?

Another example is the issue of categorizing and profiling patients based on numerous factors such as age, gender or disease. This may lead to discriminatory and exclusionary affects by insurers, physicians or hospitals. The question of how to anonymize patient information to prevent data disclosures to secondary users, such as managed care evaluators and insurance companies, all must be answered in terms of data mining. Removing personal identifiers such as name, age and social security number may make it hard to link data up to a unique individual. However, the data, while not correlated to a unique individual, may be able to be linked to a larger sub-population, such as people who live in a specific geographic region or people of a certain gender/race. The overall issue invariably becomes; which providers and organizations should get access to what data to analyze?

Another example of privacy concerns with data mining is the acceptance of payment information by clinicians. If the primary purpose for a patient to provide information at the point of care is to make a credit card payment, then any clinician providing the information for other purposes, such as data mining, without having identified this purpose with the patient before or at the time of the collection, would be considered having possibly misused patient information. The primary purpose of the data collection done by the clinician has to be clearly understood by the consumer and identified at the time of the collection. Data mining, however, is a secondary, future use. As such it requires the explicit consent of the patient. Since data mining is based on the extraction of unknown patterns from a database, a system conducting data mining does not know at the outset what personal data will be of value or what relationships will emerge from analyzing payment data. Therefore, identifying a primary purpose at the beginning of the process, and then restricting one's use of the data to that purpose, can be very challenging, especially for clinicians who are asked to protect their patient’s information from data mining.

Privacy advocates see consent, as a major requirement to promote secondary use of information, and offer three specific mechanisms:

  • Opt-out of all data mining
  • Opt-in only for use by selected secondary users
  • Opt-in for all data mining uses

Privacy concerns associated with the disclosure of information are mounting. Currently, specific legislation to deal with the privacy and security concerns of data mining has been limited. While HITECH highlights specific privacy concerns associated with secondary use of data, many government policies currently permit data mining due to public health policies in place that collect information from emergency rooms, biosurveillance and other ongoing data collection activities. Future regulatory and policy activities need to clarify and finalize the need for informed consent by patients to help clinicians deal with the ongoing issues around data mining.

Erik Pupo is health interoperability architect at Vangent Inc. in Arlington, Va. Addtionally, he currently works as a senior health IT advisor to the Office of the National Coordinator for Health Information Technology, focusing on privacy and security issues.

^ back to top

Strategies for Success

Using Performance Management to Enable and Sustain EMR Benefits

By Doug Thompson

American hospitals spent billions on electronic medical record (EMR) systems, and even greater investments are expected in the next decade.1,2,3 One reason for their popularity is the belief EMRs’ clinical decision support and workflow capabilities promote safer and more efficacious care that is also more efficient and therefore less costly. However, realizing and sustaining these benefits requires “performance management” capabilities that turn data from many hospital sources into information within a framework that promotes the intelligent use of that information.

How an EMR affects the process of care
EMRs’ decision support functions guide the decision making of clinical workers. For example, when a physician enters an order for a drug, the EMR may suggest a more cost-effective alternative. EMRs’ workflow functions support greater efficiency and safety. For example, automated clinical documentation tools collect coded data to drive decision support, improve the legibility and completeness of clinical documentation, and deliver that information to clinicians ubiquitously—saving time and preventing medical errors.

Figure 1: How an EMR affects the process of care

The role of performance management
Performance management is an emerging discipline that recently received much attention in other industries.4,5 It has two core elements:

  1. Analytic or business intelligence software that collects and aggregates data from across the organization, transforms the data into usable information, and presents it to managers and workers in a format that allows them to make better decisions and work more efficiently
  2. An organized approach to using analytics intelligently to manage individual and organizational performance

Both elements of performance management are crucial to the realization of EMR benefits. Analytic software plays three key roles, as illustrated in Figure 2:

  1. Proving to physicians and other clinicians that they need an EMR—care processes are highly variable, outcomes are not ideal. Without analytic software, it can be extremely difficult to aggregate and present the required data in an easy-to-understand format. Omitting this step hinders clinician adoption.
  2. Proving to clinicians the EMR is producing better clinical and financial results. Without results data, adoption may fade and benefits may be lost.
  3. Providing the data to ensure alerts and order sets are being used appropriately, optimize order sets, identify required modifications to clinical documentation capabilities, and guide other improvements in the EMR.

Figure 2: The role of performance management in an EMR-supported care delivery system

As Figures 1 and 2 indicate, care is delivered within individual clinicians’ mental constructs, which are affected by their education, experience, incentives and the culture of their organization. An organized and systematic approach is needed to affect these constructs and update the EMR tools that support the process of care.

A successful approach to EMR benefits realization includes:

  • Creating a “short list” of major expected benefits with detailed descriptions of how they are produced
  • Forecasting the amount of each benefit and building executive consensus around the benefits list and quantitative improvement targets
  • Creating a cross-functional executive team to lead the benefits realization effort
  • Educating physicians and other clinicians about the financial impact of their behavior
  • Aligning incentives through a formal clinician engagement strategy
  • Sharing current performance data with clinicians and addressing data concerns
  • Providing individual coaching of poor performers by clinical department leaders and one-on-one support of the medical, nursing and allied health staff during rollout
  • Continuing to share performance data; tracking and reviewing progress toward targets; using process and outcomes data to modify EMR capabilities

Without these methods, the best data in the world are likely to be ineffective in changing clinician behavior and performance. In the words of David Blumenthal, “[realizing EMR benefits is]… a change management problem, not a technology problem.”

Doug Thompson is senior vice president of performance management consulting at MedeAnalytics and a national expert in understanding, estimating, realizing and measuring health IT value.

References
1 Practice fusion. Health information technology: A bright spot in dark economic times. Available at: http://practicefusion.com/lp/arra-stimulus-plan-in-action.html.
2 Diller W. Health Care’s IT Revolution. Business Week. Feb. 12, 2007. Available at: http://businessweek.com/investor/content/feb2007/pi20070212_006800_page_2.htm.
3 Global Industry Analysts. Electronic Medical Record Systems: A North American and European Market Report. February 2010.
4 Cokins G. Performance Management: Integrating Strategy Execution, Methodologies, Risk, and Analytics. Hoboken, NJ: John Wiley & Sons; 2009.
5 Davenport T and Harris J. Competing on Analytics: The New Science of Winning. United States: Harvard Business School Press; 2007.

^ back to top

Meet an Informaticist

Alex Fiks, MD
Assistant Professor, Division of General Pediatrics
The Children’s Hospital of Philadelphia Research

 

Dr. Fiks is a practicing primary care pediatrician who realized early in his career that the treatment of individual children may be most effective when supported by systems-level interventions that address broader problems. As a result, Dr. Fiks joined the team of leaders implementing and training clinicians to use the electronic health record (EHR) at The Children’s Hospital of Philadelphia and received advanced training in epidemiology and biostatistical methods. Dr. Fik’s interests include: quality improvement in outpatient pediatrics, clinical decision support, immunization delivery and care of underserved populations.

What is the informatics professional’s role in data analysis?
I view informatics as a tool to improve the organization and delivery of healthcare.

What has been your greatest success or organizational outcome?
My work trying to improve vaccination rates for urban children illustrates the potential impact of work that encompasses a range of professionals to improve outcomes. The national benchmark for the immunization of young children is 90 percent up-to-date by 2 years of age. Through prior work and with feedback from the local health department, we recognized that children in our practices were not meeting this milestone. We formed a multidisciplinary team and sought input from front line clinicians and nurses to address this problem. The group, which involved physician informaticists, database analysts, administrators, practitioners, and biostatisticians, developed, implemented and measured the impact of point-of-care clinical alerts within the EHR that gave specific recommendations for the vaccines that were due for each child in the target age range that came to the office. Using this approach, we were able to improve vaccination rates from 78 to over 90 percent up-to-date.

What advice would you share with your peers?
As we used EHR data to achieve this system-level success, we learned many lessons. Most important is ensuring the validity of data. In our case, the validity of the immunization record depended upon the ability to capture each vaccine administered as a categorical data element. By carefully reviewing the data with our database team, we noticed that certain charted vaccines were listed as administered before birth and therefore needed to be excluded. We also noticed that sometimes doses were given too close in proximity to be considered valid based on national immunization guidelines. Further data cleansing was needed to exclude these doses.

We also learned how important it was to consider the clinical context broadly and include measures not only of the desired outcome, but of unintended consequences. In our case, we noted that improving vaccination rates by administering shots for children when they came to the office for sick visits was associated with decreases in subsequent preventive care. We explored this association because of concerns among physicians, nurses and social workers regarding this potential problem. For us, this experience demonstrated the value of collaboration between many professions in order to determine which desired and undesired consequences to consider from the start of any project. Data may be most useful if the research, reporting, or quality improvement team combines technical, clinical, and administrative expertise in assessing outcomes.

^ back to top

Special Feature

ONC Announces SHARP Awards: A Look at Secondary Use of EHR Data Research

By Christopher G. Chute, MD, DrPH

The Office of the National Coordinator for Health Information Technology (ONC) awarded $60 million in research grants through the Strategic Health IT Advanced Research Projects (SHARP) Program to the Mayo Clinic College of Medicine, Harvard University, University of Texas Health Science Center at Houston and University of Illinois at Urbana-Champaign. The Mayo Clinic College of Medicine, on behalf of a consortium of several entities, proposed research that will generate a framework of open-source services which can be dynamically configured to transform EHR data into standards-conforming, comparable information suitable for large-scale analyses, inference and integration of disparate health data.

These services will be applied to phenotype recognition (disease, risk factor, eligibility or adverse event) in medical centers and population-based settings. Finally, we examine data quality and repair strategies with real-world evaluations of their behavior in Clinical and Translational Science Awards (CTSAs), health information exchanges (HIEs) and Nationwide Health Information Network (NHIN) connections.

The Mayo-led consortium has assembled a federated informatics research community committed to open-source resources that can industrially scale to address barriers to the broad-based, facile and ethical use of EHR data for secondary purposes. We will collaborate to create, evaluate and refine informatics artifacts that advance the capacity to efficiently leverage EHR data to improve care, generate new knowledge and address population needs. Our goal is to make these artifacts available to the community of secondary EHR data users, manifest as open-source tools, services and scalable software. In addition, we have partnered with industry developers who can make these resources available with commercial deployment.

We propose to assemble modular services and agents from existing open-source software to improve the utilization of EHR data for a spectrum of use-cases and focus on three themes:

  • Normalization
  • Phenotypes
  • Data Quality/Evaluation

Our six projects span one or more of these themes, including:

  • Semantic and Syntactic Normalization;
  • Natural Language Processing (NLP)
  • Phenotype Applications
  • Performance Optimization
  • Data Quality Metrics
  • Evaluation Frameworks

Each of these services will have open-source deployments as well as commercially supported implementations.

The Sharp Program seeks to support improvements in the quality, safety, and efficiency of healthcare through advanced IT. The research projects supported by the program focus on solving current and expected challenges that represent barriers to adoption and meaningful use of health IT. See the awardees’ research focus areas below:

  • University of Illinois at Urbana-Champaign- Security of Health Information Technology
  • University of Texas Health Science Center at Houston- Patient-Centered Cognitive Support
  • Harvard University- Healthcare Application and Network Platform Architectures
  • Mayo Clinic College of Medicine- Secondary Use of EHR Data

Learn more about the SHARP Program and the Mayo Clinic College of Medicine’s progress online.

Christopher G. Chute, MD, DrPH, established the Division of Biomedical Informatics at Mayo Clinic, overseeing a program of applied research and development focusing upon clinical and genomic data sources, management, standardization and interpretation.

^ back to top

HIMSS News

HIMSS11 Call for Proposals and Reviewers Open

The Call for Proposals and the Call for Reviewers for the 2011 Annual HIMSS Conference & Exhibition—Feb. 20-24 at the Orange County Convention Center in Orlando—is open. The online proposal form will be available through May 24.

Individuals interested in submitting conference education proposals should spend time reviewing the many available online resources including information on intended audiences, various topic categories and evaluation criteria.

Reviewers and their recommendations are incredibly important to the Annual Conference Education Committee and the role that they play in the proposal selection process. Additionally, reviewers are an important resource for presentations that are selected to move forward to the annual conference; they act as coaches during the presentation preparation process and as moderators for the presentation during the conference.

This is an opportunity to make a mark on HIMSS11. Share expertise and an organization’s real-world experiences with health IT. For more information on submitting an education proposal, contact HIMSS Manager of Annual Conference Education Debra Clough at 312-915-9559.

DEA Revising Regulations on e-Prescribing Controlled Substances

The Drug Enforcement Administration (DEA) is revising its regulations to provide practitioners with the option of writing prescriptions for controlled substances electronically. The regulations will also permit pharmacies to receive, dispense and archive these electronic prescriptions. These regulations are an addition to, not a replacement of, the existing rules.

According to a summary of 334-page rule, the new rule could:

  • Help hospitals and pharmacies integrate prescription records into patients' health records;
  • Reduce paperwork for those who dispense controlled substances;
  • Reduce prescription errors due to illegible handwriting and misinterpreted oral prescriptions; and
  • Reduce prescription forgery

Comments on the Interim Final Rule are due to DEA on June 1. HIMSS will be preparing and submitting a multidisciplinary response on the IFR.

Funding Announced for the ONC Health IT Workforce Development Program

The Department of Health and Human Services awarded $84 million to 16 institutions of higher education to fund the Health IT Workforce Development Program, which focuses on several key resources required to rapidly expand the availability of health IT professionals who will support broad adoption and use of health IT in the provider community. Those resources include:

  • A community college training program to create a workforce that can facilitate the implementation and support of an electronic health care system
  • Quality educational materials that institutions of higher education can use to construct core instructional programs
  • A competency examination program to evaluate trainee knowledge and skills acquired through non-degree training programs
  • Additional university programs to support certificate and advanced degree training

The program is one of the best examples of the depth of thought behind the HITECH Act. Visit HIMSS Supporting the Health IT Workforce to learn about competencies and skill sets needed to support these efforts.

Participate in a Questionnaire about Clinical Engineering & IT Relationships

The CE-IT Community—a collaboration among AAMI, ACCE and HIMSS—is requesting participation in a questionnaire to assess the current state of working relationships between clinical engineering and IT departments in healthcare facilities.

The questionnaire addresses such issues as departmental structures, interdepartmental relationships, collaboration between clinical engineering and IT, and more. Responses will provide a clearer picture of the relationships between these critical groups. Results will be useful in identifying future projects and directions for the CE-IT Community and will be made available on the CE-IT Community Web site.

Please complete the questionnaire by Friday, April 23.

Keep HIMSS Clinical Informatics Insights coming to you! Subscribe to this complimentary e-newsletter online. Have a question, comment or idea? Please send them to Cari McLean.

Tool Box Picks

HIMSS Privacy & Security Toolkit
The toolkit outlines general principles and provides best practices and examples of how healthcare providers can manage privacy and security. Each section includes an introduction of the topic at hand; the latest edition of pertinent guidelines or literature, case studies, sample policies and Internet sources for additional information. All articles are written by experts in the field, who are at the front lines of privacy and security issues.

HISPC Provider Education Toolkit
Developed by the Health Information Security & Privacy Collaboration (HISPC), the toolkit: introduces healthcare providers to the benefits of electronic health information exchange (HIE); increases provider awareness of the privacy and security benefits and challenges of HIE; motivates providers to understand the advantages of participating in electronic HIE; identifies the steps to HIE implementation; and encourages participation in HIE.

Empirical Study on Applications of Data Mining Techniques in Healthcare
This study, by Harleen Kaur and Siri Krishan Wasan and featured in the Journal of Computer Science in 2006,examines the potential use of classification-based data mining techniques such as rule-based, decision tree and Artificial Neural Network to analyze massive volumes of healthcare data.

Data Mining Applications in Healthcare
This contribution to the Spring 2005 issue of the Journal of Healthcare Information Management authored by Hian Chye Koh and Gerald Tan, explores data mining applications in healthcare. In particular, it discusses data mining and its application within healthcare in major areas such as the evaluation of treatment effectiveness, management of healthcare, customer relationship management, and the detection of fraud and abuse.

QualityNet
This CMS-approved Web site brings together healthcare quality improvement news, resources, and data reporting tools and applications from across the Internet and organizes them in the following sections: hospitals-inpatient, hospitals-outpatient, physician offices, nursing homes, end-stage renal disease networks and facilities, and quality improvement organizations.

Developing a Data Warehouse for the Healthcare Enterprise: Lessons from the Trenches
Published by HIMSS, this book serves as a guide for healthcare executives and IT managers contemplating or involved in a data warehouse implementation. First-hand experience from individuals charged with such an implementation offers readers guidance and multiple perspectives on the data warehouse development process—from the initial vision to the system-wide release.

Join the Discussion

Check out this discussion on the HIMSS Group on LinkedIn; you have to be a member of the group to comment and read the discussions.

How does healthcare IT feel about the future sharing of data now stored in EMR silos?

Connect with HIMSS: