Fundamental Issues

On this page you will find information on several important core issues surrounding the design, implementation, legal aspects, and future goals of CDS.  This information comes from published books, papers and expert opinion on the subject; links to the original (and much more detailed) material are provided.  

Types of CDS interventions
Success Factors for CDS interventions
The CDS Five Rights
Grand Challenges in CDS
CDS and Liability

Types and examples of CDS interventions

One cannot overemphasize that CDS comes in many forms, suited for different situations and different settings. This excerpt from chapter 5 of Improving Outcomes with Clinical Decision Support: An Implementers’ Guide 1, a practical guide to developing CDS programs and implementing CDS interventions, identifies ten different types of CDS interventions. Some of these types are illustrated in the CDS Scenarios page of this website.

CDS during data-entry tasks

    • Smart Documentation Forms that are tailored based on patient data to emphasize data elements pertinent to the patient’s conditions and healthcare needs.
    • Order Sets, Care Plans and Protocols that encourage correct and efficient ordering, promote evidence-based best practices, and can provide different management recommendations for different patient situations.
    • Parameter Guidance to promote correct entry of orders and documentation.
    • Critiques and Warnings - the“Immediate Alerts” that are presented just after a user has entered an order, prescription, or documentation item, to show a potential hazard, or a recommendation for further information.

CDS during data-review tasks

  • Relevant Data Summaries (Single-patient) that summarize, filter and organize a patient’s information to highlight important management issues.
  • Multi-patient Monitors, such as a display of activity among all patients on a care unit, which help providers prioritize tasks and help ensure that important activities are not omitted while providers are multi-tasking among patients..
  • Predictive and Retrospective Analytics that combine multiple factors using statistical and/or artificial intelligence techniques to provide risk predictions, stratify patients and measure progress on broad initiatives

CDS during assessment and understanding tasks

  • Filtered Reference Information and Knowledge Resources, inclding “infobuttons” and any other place where reference information is provided in context of the current data
  • Expert Workup and Management Advisors, such as diagnostic expert systems

CDS not triggered by a user task

  • Event-driven Alerts (Data-triggered) and Reminders (Time-triggered), which alert the clinical user to a new event occurring asynchronously, such as an abnormal lab result.

This slide deck provides the types of CDS with specific examples from HIMSS Davies Award winners.

Success Factors for CDS Interventions

Common elements that are important to CDS success are summarized in the article “Ten Commandments for Effective Clinical Decision Support: Making the Practice of Evidence-based Medicine a Reality”2). This article describes the important interplay of human factors, workflow and leadership in CDS implementation. The “ten commandments” are:
A. Speed is everything
B. Anticipate needs and deliver in real time
C. Fit into the user’s workflow
D. Little things can make a big difference
E. Recognize that physicians will strongly resist stopping
F. Changing directions is easier than stopping
G. Simple interventions work best
H. Ask for additional information only when you really need it
I. Monitor impact, get feedback and respond
J. Manage and maintain your knowledge-based systems

The “CDS Five Rights”

A high level understanding of a CDS program should include an understanding of the “CDS Five Rights”. The “CDS Five Rights” summarize the ideal delivery of CDS as a mechanism for providing information that can effectively support and influence decisions. They are:
A. Right information
B. Right person
C. Right CDS intervention format
D. Right channel
E. Right point in workflow
Once a clinical target is identified, efforts should be made to identify all potential failure points that may impact the target outcome. In many instances, multiple CDS tools will be needed at different points in the clinical process. Applying different types of CDS to all of these failure points will increase the likelihood of outcome success. Both of the points above are illustrated in the flowchart.

Grand Challenges in Clinical Decision Support3

Some of the key issues in implementing a CDS system were documented in the article “Grand challenges in clinical decision support4 Through a consensus building process, the authors   identified a list of the top ten challenges to solve to foster implementable, usable, effective and widespread clinical decision support. This list was intended to help educate researchers, developers, funders, and policy-makers.

A. Improve the human-computer interface
The UI needs to support, rather than interrupt the clinical workflow. The CDS should seamless put key pieces of data and knowledge into the context of the workflow or clinical decision-making process. An improved UI could help prevent both errors of omission and commission.

B. Disseminate best practices in CDS design, development, and implementation
Some healthcare organizations have had successful experience with CDS and demonstrate  common success factors. We need to improve efforts to identify, describe, evaluate, collect, catalog, synthesize and disseminate best practices for CDS design, development, implementation, maintenance, and evaluation. Establishing methods for sharing best practices is essential for improving the effectiveness of CDS interventions and validating their performance.

C. Summarize patient-level information
The challenge for CDS systems is to intelligently and automatically analyze all of a patient’s clinical data and to create one or more brief summaries that are pertinent to the current situation. The purpose of these summaries is to make all key data needed for optimal decision-making available to each decision maker; different summaries may be needed to address the perspectives of different clinicians and workflows. In addition, information from summarization engine can be used as an input to other CDS algorithms and automatic triggering of more specific CDS becomes possible.

D. Prioritize and filter recommendations to the user
CDS systems need to appropriately account for competing influences that impact clinical decision making, and prioritize the recommendations delivered. These influences may include patient- and provider-specific factors such as: patient preferences, cost of the procedure, effectiveness of the intervention, location in the clinician’s workflow, and genetic and genomic considerations. Prioritization of the recommendations provided by the CDS system can also reduce ‘‘alert fatigue’’ that is a frequent cause of user dissatisfaction.

E. Create an architecture for sharing executable CDS modules and services
The goal is to create a set of standards-based interfaces to cloud based CDS services that any EHR could ‘‘subscribe to’’, providing access to state of the art CDS interventions with minimal effort. These services would require standardization of the definitions of and interfaces to the data required by the various CDS modules. The development of more ‘plug and play’ CDS applications could help overcome several of the key implementation barriers limiting more widespread implementation of CDS.

F. Combine recommendations for patients with co-morbidities
The challenge is to identify and eliminate redundant, contraindicated, or mutually exclusive recommendations for patients presenting with comorbid conditions or multiple medications. The CDS system should develop and present a combined recommendation from two or more clinical practice guidelines. Failure to adequately address co-morbidity or polypharmacy issues is one reason for the underutilization of these guidelines.

G. Prioritize CDS content development and implementation
Determining which CDS content to develop or implement first (e.g., patient safety, chronic disease management, or preventive health), must be based on a number of factors including patient impact, cost, availability of reliable data, difficulty of implementation, and acceptability to clinicians and patients. Replacing the current ad hoc approach to local implementations with a more, systematically prioritized approach based on national interest and overall healthcare value may lead to a much greater overall impact in the cost, safety, and quality of healthcare.

H. Create internet-accessible clinical decision support repositories
Internet-accessible repositories of high quality, evidence-based, tested, clinical decision support knowledge modules would support local deployment of content and allow local customizations, and have the ability to quickly implement on-going upgrades. Establishment of such a repository would reduce the need for each healthcare organization to develop its own rules and procedures. Some material in such repositories may come from national sources, commercial vendors, or local care delivery organizations.

I. Use free text information to drive clinical decision support
Valuable information is contained in the free text portions of EMRs.  Developing methods that would allow CDS systems to access and utilize this information would provide a new source of information. According to some reports, at least 50% of the clinical information describing a patient’s current condition and stage of therapy resides in the free text portions of the EMR.

J. Mine large clinical databases to create new CDS
There is the potential for valuable guidelines and CDS interventions to be developed and put into service, based on mining data that already exists in large clinical databases. Technical challenges exist with the creation, testing, and execution of these algorithms, as well as social challenges to insure that patient-identifiable information will remain private and secure.

CDS and Liability5

The liability issues regarding EHRs and CDS systems are too complex to be described in detail here and some issues, including the manner in which the law would be applied, are unresolved6 While there is some liability risk to providers who use CDS systems, there could also be exposure to not acting on a CDS alert. The authors of this article conclude that clinicians would be held to the same standard of care regardless of whether a CDS system is used, and as long as clinicians make the final decision the use of CDS should not increase liability risk.

Greenberg et. al. evaluated the malpractice risk associated with CDS use and concluded that the most important issue with regard to liability is whether CDS systems are well designed and well implemented.7 They determined that a well-designed CDS system should only provide alerts that are clinically relevant, which would reduce the likelihood of alert fatigue and allow clinicians to detect adverse events. This led them to conclude that adopting a well designed CDS system would reduce overall malpractice risk.

A recent article by Kesselheim et al8reached similar conclusions regarding the impact of reducing alert fatigue and emphasized the importance of using clinical judgment when interpreting the output of a CDS system. To further reduce liability, they also recommend stronger government regulation of clinical decision support systems and the development of international practice guidelines.

The issue of sharing CDS content between institutions is discussed by Wright et al9 in the context of using a Web 2.0 architecture to encourage interoperability. Citing other references, they conclude that the patient’s health care provider is responsible for making the final decision on the clinical relevance of any shared CDS content. He authors state that this model is similar to the use of existing “hard copy” references to aid in clinical decision making. However, they caution that there is little precedent to definitively answer these questions.


1Osheroff JA, Teich JM, Levick D, Saldana L, Velasco FT, Sittig DF, Rogers KM, Jenders RA. Improving Outcomes with Clinical Decision Support: An Implementer’s Guide. HIMSS press, 2012.

2Bates DW, Kuperman GJ, Wang S, Gandhi T, Kittler A, Volk L, Spurr C, Khorasani R, Tanasijevic M, Middleton B. Ten commandments for effective clinical decision support: Making the practice of evidence-based medicine a reality. JAMIA 2003;10:523-530.

3Sittig DF, Wright A, Osheroff JA, Middleton B, Teich JM, Ash JS, Campbell E, Bates DW. Grand challenges in clinical decision support. J Biomed Inform. 2008 Apr;41(2):387-92

4 Ibid

5Kesselheim A. S., Cresswell K., Phansalkar S., Bates D. W., Sheikh A.Clinical decision support systems could be modified to reduce ‘alert fatigue’ while still minimizing the risk of litigation. Health Aff2011; 30: 2310-7

6Miller RA, Miller SM. Legal and regulatory issues related to the use of clinical software in health care delivery. In: Greenes RA, editor. Clinical decision support: the road ahead. Amsterdam: Elsevier Academic Press; 2007. p. 423-44

7Greenberg, M. and M. S. Ridgely (2011). Clinical decision support and malpractice risk JAMA: The Journal of the American Medical Association 306 (1), 90-91.

8 Kesselheim A. S., Cresswell K., Phansalkar S., Bates D. W., Sheikh A.Clinical decision support systems could be modified to reduce ‘alert fatigue’ while still minimizing the risk of litigation. Health Aff2011; 30: 2310-7Ibid,

9Wright, A, et al, Creating and sharing clinical decision support content with Web 2.0: Issues and example., Journal of Biomedical Informatics 42 (2009) 334-346