Chapter 10: Incorporating Evidence: Use of Computer-based Clinical Decision Support Systems for Health Professionals
Introduction Decision Support Systems (DSS) - automated tools designed to support decision-making activities and improve the decision-making process and decision outcomes - Intended to use the enormous amounts of data that exist in information systems to facilitate decision processes
Clinical Decision Support Systems (CDSS) - Set of knowledge-based tools that can be fully integrated with the clinical data embedded in the computerized patient record (electronic health record) to assist providers by presenting information relevant to the healthcare problems being faced - Only as effective as its underlying knowledge base - Tool system not a rule system
GOALS: Patient safety Improved outcomes for specific patient populations Compliance with clinical guidelines Standards of practice Regulatory requirements Primary Goal: Optimization of both the efficiency and effectiveness with which clinical decisions are made and care is delivered
Nursing Decision Support Systems (NDSS) – tools that help nurses improve their effectiveness, identify appropriate interventions, determine areas in need of policy or protocol development, and support patient safety initiatives and quality improvement activities
Definition CDSS – any computer program that helps health professionals make clinical decisions Johnston – computer software employing a knowledge base designed for use by a clinician involved in patient care, as a direct aid to clinical decision-making Sims – software designed to be a direct aid to clinical decision-making , in which characteristics of an individual patient are matched to a computerized clinical knowledge base and patient-specific assessments or recommendations are then presented to the clinician or the patient for a decision
Coiera – role of CDSS – augmenting human performance and providing assistance for healthcare providers Berner - healthcare is being transformed through information and knowledge management and technology is being used to “tame data and transform information
Expanded Uses of CDSS Randall Tobias – former VP of ATT - Computer has virtually unlimited capacity for processing and storage of data - Human has limited storage(memory) and processing power, but does have judgment, experience, and intuition.
THREE main purposes of a DSS: 1. Assist in problem solving with semi structured problems 2. Support, not replace, the judgment of a manager or clinician 3. Improve the effectiveness of the decision- making process
History of CDSS Early Systems Focus on Diagnosis Earliest known CDSS developed by de Dombal in 1972 at Leeds University - Used Bayesian theory to predict the probability that a given patient, based on symptoms, had one of seven possible conditions 1974 – INTERNIST I – developed at the University of Pittsburgh to support the diagnostic process in general internal medicine by linking diseases with symptoms - Later became the basis of successor systems including quick medical reference (QMR)
1976 – MYCIN – rule-based expert system to diagnose and recommend treatment for certain blood infections In nursing, TWO early and well known systems 1. COMMES (Creighton online multiple modular expert systems) 2. CANDI (computer aided nursing diagnosis and intervention - These were developed to assist nurses with care planning and nursing diagnosis
Types and Characteristics of DSS Types of DSS Administrative and Organizational Systems Administrative Systems (financial or quality monitoring)- generally support the business decision- making process; encompass decision processes - Tend to be batch-oriented Focused on: Real-time decision support Goal orientation Intelligence gathering designed to be used at the point of care by clinicians
Integrated Systems – support outcomes performance management by integrating operational data: Budgeting Executive decision-making Financial analysis Quality management Strategic planning data
Characteristics of DSS Shortliffe – uses function, mode of advice, consultation style, underlying decision-science methodology, and user- computer interactions to categorize systems Teich and Wrinn – examine DSS from the aspects of functional and logical classes and structural elements
Perreault – organized key CDSS functions as: 1. Administrative – support for clinical coding and documentation 2. Management of clinical complexity and details – keeping patients on research and chemotherapy protocols, tracking orders, referrals, follow-up, and preventive care 3. Cost control – monitoring medication orders and avoiding duplicate or unnecessary tests 4. Decision support – supporting clinical diagnostic and treatment plan processes promotion of best practices, use of condition-specific guidelines, and population- based management
DSSs could be divided into: a. Data-based(population-based) b. Model-based(case-based) c. Knowledge-based(rule-based) d. Graphics-based systems
A. Data-based systems - Provide decision support - Capitalize on the fundamental input into any intelligent system, data - Provide decision support OLAP – on line analytic processing
B. Model-based DSSs – driven by access to and manipulation of a statistical, financial, optimization, and/or simulation model Model – generalization that can be used to describe the relationships among a number of observations to represent a perception of how things fit together Genetic algorithms (GAs) and neural networks (NNs) – newer computation techniques that are evolving as problem solving solutions
C. Knowledge-based systems – rely on expert knowledge that is either embedded in the system or accessible from another source and uses some type of knowledge acquisition process to understand and capture the cognitive processes of healthcare providers EBP – evidence-based practice
D. Graphics-based systems – take advantage of the user interface to support decisions by providing decision “cues” to the user in the form of color, graphical representation options, and data visualization
**Demand management centers use decision tree logic (DTL) or rule-based logic (RBL) for patient management. DTL – useful for specific straightforward tasks RBL – allows for complex decision capacities - More flexible with answers, provides consistent outcomes, and is adaptable to change - Also tends to have rigid solutions and allows little or no clinician autonomy
Taxonomy for CDSS: 1. Context 2. Knowledge and data sources 3. Workflow 4. Decision support 5. Information delivery Institute of Medicine (IOM) – human error as a major source of patient care morbidity and mortality
Barriers to the Use of CDSS Systems Bates – practice lags behind knowledge by several years - This lag could be shortened if not eliminated by the availability of current knowledge to support the decision-making process
Henry – identified essential elements needed for an informatics infrastructure: 1. Standardized vocabularies to describe patient diagnoses, interventions, and outcomes 2. Computer-based methods to examine linkages among patient problems and characteristics, healthcare interventions, patient outcomes, and the intensity of care-resources and to examine practice variations 3. An integrated clinical information system where data required for quality improvement are both collected and returned to the provider during routine processes of patient care
Evaluation of CDSS Sittig – cites the following FIVE elements: 1. Integrated real-time patient database – which combines patient data from multiple sources, lab, radiology, pharmacy, admissions, nursing notes, and so on; provide context for results interpretation 2. Data-drive mechanism – allows even triggers to go into effect and activate alerts and reminders automatically 3. Knowledge engineer – translate the knowledge representation scheme used in system so that the clinical knowledge in the system can be extracted and translated into the machine executable logic
4. Time-driven mechanism – to permit automatic execution of programs at a specific time to alert provider to carry out a specific action or insure that the action had been completed 5. Long-term clinical data repository – data collected over time from a variety of sources allowing a longitudinal patient record
Knowledge and Cognitive processes Knowledge engineering – field concerned with knowledge acquisition and the organization and structure of that knowledge within a computer system Interviews – MOST COMMONLY used method of eliciting knowledge
Cognitive task analysis (CTA) – set of methods that attempt to capture the skills, knowledge, and processing ability of experts in dealing with complex tasks Attempts to identify pitfalls or trouble spots in the reasoning process of the beginner or intermediate level practitioner GOAL of CTA: Tap into these “higher order” cognitive functions
Tan and Sheps – six-step approach to CTA: 1. Identification of the problem to target in the analysis 2. Generation of cases (decision tasks) that vary on key factors 3. Observation of a record of an expert problem solving for the case using think aloud 4. Observation of the novice and the intermediate problem solving 5. Analyses of expert versus less than expert problem solving 6. Recommendation of systems needs, design specs, and knowledge base components
Responsibility of User: Ethical and Legal Issues
Implications for Future Uses of CDSS in Nursing Increasing Inclusion of Patients CDSS allow patient access to the knowledge base of the system The computer can become a patient health medium with reference databases, library access for healthcare information, drug and disease management information, self-help programs, and advice about prevention available
Dual Purpose of Documentation - Balance the use of poorly designed or inadequately tested systems with individual clinicians being forced to make patient care decision-making without existing evidence at the point of care DUAL Purpose: 1. Improving care for the individual patient 2. Improving care for future populations of patients via aggregated information used for clinical decision-making
CDSS can: Improve patient care quality Reduce medication errors Minimize variances in care Improve guideline compliance Promote cost savings