2
LAYING THE CDS FOUNDATION STONES

This project draws from and builds upon important initiatives to improve the delivery of care through effective CDS. In essence, CDS is the twenty-first century version of clinical practice guidelines (CPGs)—systematically developed statements to assist practitioner and patient decisions about appropriate health care for specific clinical circumstances (Institute of Medicine, 1990)—presented in real time within the context of a patient’s EHR. In 2011, a report by the Institute of Medicine (now the National Academy of Medicine), Clinical Practice Guidelines We Can Trust stated that clinicians can no longer stay abreast of the rapidly expanding knowledge base related to medicine, that “clinicians increasingly are barraged with a vast volume of evidence of uncertain value,” and that increased adoption of EHRs and CDS offers unique opportunities to rapidly move clinical knowledge from the scientific literature to the patient encounter. At the same time, the report noted, CPGs—and by inference CDS since it is often based on CPGs—do not dictate a one-size-fits-all approach to patient care but rather serve to enhance clinician and patient decision-making by clearly describing and appraising the scientific evidence behind clinical recommendations and making them relevant to the individual patient encounter (Institute of Medicine, 2011).

This chapter highlights contributions to developing strategies to develop CDS as an important component of EHR systems. These prior efforts have informed this project and provided important lessons for this current initiative.

A ROADMAP FOR NATIONAL ACTION ON CDS

In 2005, ONC commissioned the American Medical Informatics Association (AMIA) to develop a tactical plan to guide federal and private sector activities to advance the development and adoption of CDS. The resulting roadmap, issued in 2006, included three pillars and six strategic objectives for CDS to “ensure that optimal, usable, and effective clinical decision support is widely available to providers, patients, and individuals where and when they need it to make health

care decisions” (Osheroff et al., 2007). The AMIA roadmap’s three pillars and six strategic objectives were:

• Pillar 1: Best knowledge available when needed
  • Strategic Objective A: Represent clinical knowledge and CDS interventions in standardized formats so that a variety of knowledge developers can produce this information in a way that knowledge users can readily understand, assess, and apply it.
  • Strategic Objective B: Collect, organize, and distribute clinical knowledge and CDS interventions in one or more services from which users can readily find the specific material they need and incorporate it into their own information systems and processes.
• Pillar 2: High adoption and effective use
  • Strategic Objective C: Address policy/legal/financial barriers and create additional support and enablers for widespread CDS adoption and deployment.
  • Strategic Objective D: Improve clinical adoption and usage of CDS interventions by helping clinical knowledge and information system producers and implementers design CDS systems that are easy to deploy and use, and by identifying and disseminating best practices for CDS deployment.
• Pillar 3: Continuous improvement of knowledge and CDS methods
  • Strategic Objective E: Assess and refine the national experience with CDS by systematically capturing, organizing, and examining existing deployments. Share lessons learned and use them to continually enhance implementation best practices.
  • Strategic Objective F: Advance care-guiding knowledge by fully leveraging the data available in interoperable EHRs to enhance clinical knowledge and improve health management.

The roadmap included a comprehensive work plan that outlined the full set of tasks needed to create both a robust infrastructure for developing and delivering CDS interventions and an environment that encourages widespread adoption and continual refinement of these interventions. It also included a set of critical path tasks that could be implemented and produce results in the near term, and provide a foundation for further efforts to create a national CDS infrastructure, as well as a straw-man proposal for demonstrating a scalable, outcome-enhancing CDS system.

Lessons from the roadmap exercise included the need to develop standard formats for knowledge and interventions and to conceive approaches for collecting and distributing CDS. The roadmap process also identified legal and financial barriers that needed to be addressed and determined that mechanisms were needed to compile and disseminate best practices for usability and implementation and to improve CDS through actual experience and by mining EHR data systematically to advance knowledge. To foster action on elements of the Roadmap, ONC has also funded related work exploring CDS implementation (Advancing CDS)² and CDS standards harmonization (Health eDecisions and the Clinical Quality Information Workgroup).^3,4

THE CDS CONSORTIUM

In 2008, AHRQ funded a five-year project, the Clinical Decision Support Consortium (CDSC) (Middleton, 2009) to assess, define, demonstrate, and evaluate best practices for knowledge management and CDS in health IT across multiple ambulatory care settings EHR technology platforms. Members of the CDSC included academic and community provider institutions, leading health IT organizations, EHR companies, and knowledge vendors from across the nation. The CDSC solved critical technical challenges for sharing CDS (Boxwala et al., 2011) and developing social and legal frameworks that facilitate such sharing (Wright et al., 2011). The project selected a service-oriented approach to providing clinical decision support (Sittig et al., 2009). Web services were developed at Brigham and Women’s Hospital, the lead CDSC site, and made available to consumers across the United States (Wright et al., 2009). In addition, both human and machine-readable artifacts were made available.

STRUCTURING CARE RECOMMENDATIONS FOR CDS

From 2009 to 2011, AHRQ funded a project to develop a process for translating narrative, unstructured, evidence-based clinical recommendations and performance measures into a structured, coded format that can be implemented into health IT systems, applications, and products. The goal for developing such a

process was to enable local health IT systems to more easily integrate robust CDS rules into local health IT systems, potentially broadening adoption of CDS and leading to improved patient care and outcomes. These structured recommendations, developed for all 50 of the U.S. Preventive Services Task Force A and B recommendations⁵ and all 12 recommendations relevant to meaningful use measures that must be reported to the Centers for Medicare and Medicaid Services, became known as eRecommendations (Raetzman et al., 2011). These eRecommendations leverage standard data elements, coding systems, and value sets developed for performance reporting under meaningful use for health IT to identify patients for whom a clinical recommendation applies and action should be taken. Throughout the project, the format and content of eRecommendations were vetted extensively with multiple stakeholders. Broad stakeholder feedback, which included health care provider organizations, guideline developers, EHR, and CDS suppliers, indicated wide interest in the eRecommendation work and belief that the project materials could deliver significant value. CDS needs to be specifically tested in an electronic environment, as paper-based systems invariably require some degree of judgment in application, whereas CDS, by definition, is triggered not by judgment, but by data.

CLINICAL PRACTICE GUIDELINES

The 2011 IOM report, Clinical Practice Guidelines We Can Trust, discussed some of the evidence showing the benefits of CDS, but also noted the existence of a few studies offering contrasting results. One such study, for example, found that CDS designed to improve diabetes and coronary artery disease care among primary care physicians resulted in limited effectiveness. Although reminders increased the odds that participants followed recommended care, adherence to quality measures remained low and significant variability in practice persisted (Sequist et al., 2005). A 2004 evaluation of a guideline-based computerized educational tool found no significant difference in guideline knowledge between physician groups with and without access to the tool (Butzlaff et al., 2004), while a 2002 study of CDS to aid implementation of CPGs for the management of asthma and angina by primary care practitioners found that CDS had no significant effect on consultation rates, process of care measures including prescribing, as structured for that program, or any patient reported outcomes for either condition (Eccles et al., 2002).

One of the limitations that existed at the time of the 2011 report was that even basic EHRs—those with the ability to record patient demographic and health data and manage prescription order entry, laboratory, and imaging results—let alone those with CDS capabilities, were scarce. One study summarizing the evidence related to EHRs reported that the quality of the data on hospital EHR adoption is generally poor, and that, at the time, only approximately five percent of hospitals had computerized physician order entry, which is just one crucial element of EHRs. (Jha et al., 2006). The situation was slightly better in ambulatory care settings, with some 17 percent of ambulatory care clinics having basic EHR capabilities and 4 percent using a comprehensive EHR (DesRoches et al., 2008). The HITECH Act, the associated stages of satisfying meaningful use criteria, and accompanying financial incentives have largely addressed that shortcoming.

Moreover, since CDS implementation has been advancing in a number of places under different circumstances, a new evaluation environment has emerged. For example, the Veterans Health Administration has had CDS in place for more than a decade; it may be an environment for determining some of the benefits and challenges relevant to accelerating effective CDS more broadly.

HIMSS CDS101

The Healthcare Information and Management Systems Society (HIMSS) developed CDS101 to provide a broad and concise overview of CDS, including implementation challenges and strategies to overcome them, for health care organizations interested in implementing CDS within their health IT infrastructure. CDS101 includes a downloadable,⁶ customizable C-Suite level presentation that outlines the challenges and leadership commitment needed to ensure a successful CDS program, and it provides detailed discussions of the promise and perils of CDS adoption. The CDS101 program provides a range of scenarios for how CDS is deployed in various health care environments and a toolbox that describes the types of CDS interventions and success factors for CDS interventions. The toolbox lists what HIMSS calls the “CDS Five Rights”:

1. Right information
2. Right person
3. Right CDS intervention format

4. Right channel
5. Right point in workflow

In creating CDS101, HIMSS paid particular attention to discussing what have been called the grand challenges in CDS (Sittig et al., 2008), which are to:

• Improve the human-computer interface to support rather than interrupt the clinical workflow.
• Disseminate best practices in CDS design, development, and implementation.
• Summarize patient-level information intelligently and automatically to create one or more brief summaries pertinent to the current situation and that can provide all key data needed for optimal decision making.
• Prioritize and filter recommendations to the user in a manner that accounts for the competing influences that impact clinical decision making and prioritizes delivered recommendations.
• Create an architecture for sharing executable, plug-and-play CDS modules and services in a cloud-based environment to which any EHR system could “subscribe” without the need to recreate clinically proven CDS.
• Prioritize CDS content development and implementation, rather than continue doing so on an ad hoc basis, based on a number of factors, including patient impact, cost, availability of reliable data, implementation difficulty, and acceptability to clinicians.
• Create internet-accessible CDS repositories of high-quality, evidence-based, tested CDS knowledge modules that would support local deployment of content, allow for local customizations, and enable rapid upgrades of context with the development of new knowledge (Sittig et al., 2009). Establishing this repository would reduce the need for every health care organization to develop its own rules and procedures.
• Use free-text information to harvest valuable information in EHRs to drive CDS.
• Mine existing large clinical databases to create new CDS.

AHRQ PCOR CDS INITIATIVE

Building on its long history of investments to advance CDS, including the CDSC and GLIDES project (GuideLines Into Decision Support), AHRQ established the Patient-Centered Outcomes Research (PCOR) CDS Initiative in 2016 to promote the dissemination and implementation of PCOR CDS findings and develop tools to help CDS become more shareable, health IT

standards-based, and publicly available and to create reusable CDS modules and tools and a CDS repository.⁷ According to AHRQ, PCOR-based CDS helps patients and their care teams apply evidence from patient-centered outcomes research to enhance care processes and their results. Approaches include promoting shared decision-making, incorporating patient reported outcomes, factoring in patient preferences to generate patient-specific recommendations for care, and others. This initiative will have four main components: PCOR CDS Learning Network, CDS Connect, two funding opportunities to scale existing CDS and develop new CDS, and an evaluation effort for the overall initiative.

The PCOR CDS Learning Network, based at RTI International, is building a community of researchers, clinicians, professional societies, and others to accelerate collaborative learning opportunities and advance patient-centered CDS. The Learning Network will create content and a collaboration hub containing information that promotes understanding of patient-centered CDS, disseminates patient-centered evidence and practice, disseminates best practices for incorporating evidence into patient-centered CDS, and shares information on approaches to dissemination, development, implementation, and evaluation of patient-centered CDS. The Learning Network’s stakeholders will contribute to the creation of relevant technical standards, policies, legal frameworks, and market analyses aimed at creating momentum for widespread adoption of patient-centered CDS. The strategic foci of the Learning Network will be to provide stakeholders with a broad array of up-to-date information relevant to patient-centered CDS, to provide information and services that enable stakeholders to connect and collaborate, and to foster the collaborative development and application of concepts, frameworks, policies, and standards for patient-centered CDS while recognizing that not all PCOR findings are suitable for implementation in CDS. A key concept underlying this work is that, at a minimum, patient-centered CDS includes an element of patient choice, whether direct or by proxy.

One of the first activities of the PCOR CDS Learning Network was to identify barriers and facilitators to the dissemination of PCOR-based CDS (Richardson et al., 2016). A critical artifact that grew out of this effort is the Analytic Framework for Action (AFA) (Figure 2–1). The AFA provides a means by which the CDS community can organize the findings and recommendations of the PCOR CDS Learning Network, and it represents the lifecycle of activities that must occur to disseminate PCOR through CDS, measure its impact,

and create a learning system. Throughout the process of prioritizing evidence for dissemination via CDS, authoring CDS interventions, implementing those interventions, measuring the decisions and outcomes from those interventions, and learning from the CDS experience at each step, it will be important to recognize and manage external factors, such as the marketplace, policy, legal, and governance factors that affect developing, dissemination, and implementation processes for patient-centered CDS.

CDS Connect, led by MITRE, will demonstrate a web-based repository service that will enable the broader PCOR CDS community to identify evidence-based standards of care, provide a tool to promote a collaborative model of CDS development, and translate and codify information into an interoperable standard. The repository will offer structured data, aggregated resources, and the ability to leverage the international standard Clinical Quality Language. As a demonstration, CDS Connect is focusing on CDS related to cholesterol management.

SUMMARY OF KEY ISSUES

As Jonathan Teich noted in his presentation to the first workshop, these and other efforts have shown that to be useful and accepted CDS needs to be several things. “It needs to be smart. It needs to be aware of the context and be like the guru down the street that can actually give you answers,” said Teich. “It needs to be filtered and sensitive to the patient.” CDS, he added, needs to provide alerts that are useable within the workflow so that the user’s experience is clean and easy, and it needs to be shareable, valuable, safe, and perhaps, above all, CDS must become an important part of a learning health system.

In a recent review of the field, Blackford Middleton and colleagues noted that CDS has evolved dramatically over the past 25 years and will likely evolve just as dramatically or more so over the next 25 years (Middleton et al., 2016). They suggest that this evolution is inevitable given the explosion of biomedical knowledge and the pressure to improve the quality of care and lower costs in value-based care. While the projects described above, as well as others, have made significant progress in demonstrating how to develop effective CDS for specific cases, widespread adoption of CDS to improve care has not occurred. As numerous speakers at the three workshops noted, there are multiple challenges that the field still needs to address to realize this vision, including those involving authoring CDS, the technical implementation of CDS, operations, and scaling and spreading the value proposition. These challenges will be discussed more fully in the next chapter.

REFERENCES

Eccles, M., E. McColl, N. Steen, N. Rousseau, J. Grimshaw, D. Parkin, and I. Purves. 2002. Effect of computerised evidence based guidelines on management of asthma and angina in adults in primary care: Cluster randomised controlled trial. BMJ 325(7370):941.

Institute of Medicine. 1990. Clinical practice guidelines: Directions for a new program. Edited by M. J. Field and K. N. Lohr. Washington, DC: The National Academies Press.

———. 2011. Clinical practice guidelines we can trust. Edited by R. Graham, M. Mancher, D. M. Wolman, S. Greenfield and E. Steinberg. Washington, DC: The National Academies Press.

Jha, A. K., T. G. Ferris, K. Donelan, C. DesRoches, A. Shields, S. Rosenbaum, and D. Blumenthal. 2006. How common are electronic health records in the United States? A summary of the evidence. Health Aff (Millwood) 25(6):w496-507.

Middleton, B. 2009. The clinical decision support consortium. Stud Health Technol Inform 150:26–30.

Middleton, B., D. F. Sittig, and A. Wright. 2016. Clinical decision support: A 25 year retrospective and a 25 year vision. Yearb Med Inform Suppl 1:S103-116.

Osheroff, J. A., J. M. Teich, B. Middleton, E. B. Steen, A. Wright, and D. E. Detmer. 2007. A roadmap for national action on clinical decision support. J Am Med Inform Assoc 14(2):141–145.

Raetzman, S. O., J. Osheroff, R. A. Greenes, et al. Structuring Care Recommendations for Clinical Decision Support: Final Report. (Prepared by Thomson Reuters under Contract No. HHSA 290–2009-00022I.) AHRQ Publication No. 11–0025–2-EF. Rockville, MD: Agency for Healthcare.

Research and Quality. September 2011. Richardson, J. E., B. Middleton, J. A. Osheroff, M. Callaham, L. Marcial, and B. H. Blumenthal. 2016. The PCOR CDS-LN environmental scan: Spurring action by identifying barriers and facilitators to the dissemination of PCOR through PCOR-based clinical decision support. Research Triangle Park, NC: RTI International.

Sequist, T. D., T. K. Gandhi, A. S. Karson, J. M. Fiskio, D. Bugbee, M. Sperling, E. F. Cook, E. J. Orav, D. G. Fairchild, and D. W. Bates. 2005. A randomized trial of electronic clinical reminders to improve quality of care for diabetes and coronary artery disease. J Am Med Inform Assoc 12(4):431–437.

Sittig, D. F., A. Wright, J. S. Ash, and B. Middleton. 2009. A set of preliminary standards recommended for achieving a national repository of clinical decision support interventions. AMIA Annu Symp Proc 2009:614–618.

Sittig, D. F., A. Wright, J. A. Osheroff, B. Middleton, J. M. Teich, J. S. Ash, E. Campbell, and D. W. Bates. 2008. Grand challenges in clinical decision support. J Biomed Inform 41(2):387–392.

Wright, A., D. W. Bates, B. Middleton, T. Hongsermeier, V. Kashyap, S. M. Thomas, and D. F. Sittig. 2009. Creating and sharing clinical decision support content with web 2.0: Issues and examples. J Biomed Inform 42(2):334–346.

Wright, A., D. F. Sittig, J. S. Ash, D. W. Bates, J. Feblowitz, G. Fraser, S. M. Maviglia, C. McMullen, W. P. Nichol, J. E. Pang, J. Starmer, and B. Middleton. 2011. Governance for clinical decision support: Case studies and recommended practices from leading institutions. J Am Med Inform Assoc 18(2):187–194.

This page intentionally left blank.