Using Commercial Knowledge Bases for Clinical Decision Support: Opportunities, Hurdles, and Recommendations

From Clinfowiki
Jump to: navigation, search

These are reviews for Gilad J. Kuperman, Richard M. Reichley, and Thomas C. Bailey's 2006 article, Using commercial knowledge bases for clinical decision support: opportunities, hurdles, and recommendations.

Review 1

Using e-prescribing as focus and example, this editorial commentary describes the opportunities and hurdles in using commercial knowledge bases for clinical decision support systems, and makes recommendations for knowledge base vendors and clinical information system (CIS) developers in working towards realizing the benefits of clinical decision support (CDS) and accelerating the availability of clinically useful knowledge bases. Two seemingly contradicting words flash in my mind as I read this article: customization and standardization.

This review is useful for clinical information system developers, information technology professionals and administrators who are concerned about implementing improving health care quality and safety.

Introduction

In this article, the authors assert that “the quality and safety of health care leaves much to be desired” by citing two published reports on quality of health care. [1] [1,2] Given the complexity of health care system and barriers to implementing the practices and policies needed to improve safety of patient care, using commercial knowledge bases for CDS is one way towards improving quality and safety of health care.

Opportunities

Reports of studies have shown that computerized medication ordering process using pharmaceutical knowledge bases can reduce medication errors. [3] However, most organizations do not have the expertise or resources to crate such knowledge bases themselves. The solution to this problem of access to medication-related knowledge is to buy a commercially produced knowledge base that contain drug-drug, drug-disease interactions, minimum and maximum dosing suggestions, drug-allergy cross-sensitivity groups, and groupings of medications by therapeutic class.

Hurdles

CDS with commercial knowledge bases are not well received because they generate excessive number of alerts. This causes clinicians to have decreased confidence in CDS and to ignore all alerts (i.e. the unhelpful, nonsense, overly sensitive alerts as well as clinically relevant ones). The authors fear that this decreased confidence in the alerting system and CDS can lead to clinicians’ dissatisfaction with clinical information system and impede the nation’s progress towards the goals of the health IT strategic framework set forth by NIH.[4]

Recommendations

The authors of this article recommend that developers of clinical information systems and vendors of knowledge bases work together to:

  1. Design knowledge bases that are customizable, modifiable, and browsable
  2. Enable the local customizations be retained and not be affected by the updates
  3. Describe for the users how the knowledge base and the clinical system interact
  4. Make the customizations made by one user can be exported to another user
  5. Adopt standards for knowledge base representation and concept identifiers.
  6. Facilitate sharing and evaluation of customization efforts and results.

Review 2

Medication ordering and management is an aspect of healthcare delivery that is plagued with quality and safety issues. These issues could be alleviated by medication-related clinical decision support (CDS) embedded in Computerized Provider Order Entry (CPOE) or Electronic Health Record (EHR) systems.

Representing medical knowledge in information systems has been a challenge but this knowledge acquisition bottleneck can be eased by utilizing knowledge bases developed by commercial vendors. Such knowledge bases are cost-effective and can be incorporated into Clinical Information Systems (CIS) to display alerts highlighting drug-drug and drug-disease interactions, inappropriate medication doses, drug allergies, and duplicate medication orders.

In spite of their apparent benefits, such knowledge bases have not gained wide acceptance mainly due to their propensity for excessive and clinically unhelpful alerts. Such nuisance alerts not only degrade a clinician’s performance but also his confidence in the alerts and ultimately lead to dissatisfaction with the clinical information system.

Allowing healthcare organizations to customize behavior of CIS by filtering unnecessary alerts can improve the value of alert functionality of a CIS. But healthcare organizations face several challenges in this effort:

  • Lack of management processes to decide on customizations to knowledge-base or CIS functionality
  • Closed architecture for CIS and knowledge base
  • Unsustainability of customizations as they may be overwritten by subsequent CIS or knowledge-base upgrades
  • Inability to share customizations across multiple healthcare entities
  • Liability concerns regarding inadvertent suppression of critical alerts
  • Complications in legal agreements between healthcare organizations, CIS, and knowledge-base vendor.

To meet these challenges, the authors make several recommendations:

  • CIS and knowledge base vendors should provide tools to allow end-user customizations for filtering clinically unhelpful alerts or supplementing it with additional checks.
  • Vendors should develop better tools to browse the knowledge-base
  • Vendors should provide information about system architecture to facilitate customizations
  • Product design should allow sharing of customizations across organizations
  • Standards should be developed and adopted for knowledge base representation
  • HCOs should implement specific policies and procedures regarding knowledge editing

Comments

Allowing healthcare organizations to customize their medication-related knowledge base and clinical information systems would help align the CIS to organizational preferences. It would also allow clinicians to control the system behavior and filter out clinically unimportant alert. Only when the knowledge bases and clinical information systems support and add value to healthcare delivery, the full potential of CDS systems can be realized.


Using Commercial Knowledge Bases for Clinical Decision Support: Opportunities, Hurdles, and Recommendations. J Am Med Inform Assoc. 2006;13:369 –371


References

  1. Gilad J. Kuperman, Richard M. Reichley, and Thomas C. Bailey. Using Commercial Knowledge Bases for Clinical Decision Support: Opportunities, Hurdles, and Recommendations. J Am Med Inform Assoc. 2006 Jul-Aug; 13(4): 369–371. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1513681/


  1. McGlynn EA, Asch SM, Adams J, Keesey J, Hicks J, DeCristofaro A, Kerr EA. The quality of health care delivered to adults in the United States. N Engl J Med. 2003 Jun 26, 348(26):2635-45.
  2. Leape LL, Berwick DM. Five years after To Err Is Human: what have we learned: JAMA. 2005 May 18; 293(19): 2384-90.
  3. Hunt DL, Haynes RB, Hanna SE, Smith K. Effects of computer-based clinical decision support systems on physician performance and patient outcomes: a systematic review. JAMA. 1998;280:1339-46.
  4. Department of Health and Human Services. Goals of the Health Information ?Technology Strategic Framework. http://www.os.dhhs.gov/healthit/goals.html. Accessed May 18, 2007.

Beshia Popescu