Clinical Decision Support Notes

Chapter 8: Clinical Decision Support

Learning Objectives

  • Define electronic clinical decision support (CDS).
  • Enumerate the goals and potential benefits of CDS.
  • Discuss the government and private organizations supporting CDS.
  • Discuss CDS taxonomy, functionality, and interoperability.
  • List the challenges associated with CDS.
  • Enumerate CDS implementation steps and lessons learned.

Introduction

  • Definition: Clinical decision support (CDS) provides clinicians, staff, patients, or other individuals with knowledge and person-specific information, intelligently filtered or presented at appropriate times, to enhance health and health care (ONC).
  • Any resource that aids in decision-making should be considered CDS, but the focus is on electronic CDS.
  • Clinical decision support systems (CDSSs) are the technology that supports CDS.
  • Early CDS was primarily reminders and alerts, but now includes diagnostic help, cost reminders, calculators, etc.
  • Many organizations promote CDS to improve patient safety.
  • Most CDS strategies involve the 5 rights:
    • The right information (what): based on the highest level of evidence possible and adequately referenced.
    • To the right person (who): the person making the clinical decision (physician, patient, or other team member).
    • In the right format (how): alert, reminder, infobutton, or order set.
    • Through the right channel (where): EHR alert, text message, email alert, etc.
    • At the right time (when): early in the order entry process.

Historical Perspective

  • As early as the 1950s, scientists predicted computers would aid medical decision-making.
  • CDS programs appeared in the 1970s but were standalone and eventually became inactive.
  • De Dombal’s system for acute abdominal pain: used Bayes theorem to suggest differential diagnoses.
  • Internist-1: CDS program that used IF-THEN statements to predict diagnoses.
  • Mycin: rule-based system to suggest diagnosis and treatment of infections.
  • DxPlain: 1984 program that used clinical findings to list possible diagnoses (now a commercial product).
  • QMR: began as Internist-1 for diagnoses and ended in 2001.
  • HELP: began in the 1980s at the University of Utah and includes diagnostic advice, references, and clinical practice guidelines.
  • Iliad: diagnostic program developed by the University of Utah in the 1980s.
  • Isabel: commercial differential diagnosis tool with free text input from EHR; uses natural language processing and supported by 100,000 documents.
  • SimulConsult: diagnostic program based on Bayes probabilities; predictions include clinical and genetic information.
  • SnapDx: free mobile app that performs diagnostic CDS for clinicians, based on positive and negative likelihood ratios from medical literature; covers about 50 common medical scenarios.

CDS Benefits and Goals

  • Improvement in patient safety:
    • Medication alerts.
    • Improved ordering.
  • Improvement in patient care:
    • Improved patient outcomes.
    • Better chronic disease management.
    • Alerts for critical lab values, drug interactions, and allergies.
    • Improved quality-adjusted life years (QALY).
  • Reduction in healthcare costs:
    • Fewer duplicate lab tests and images.
    • Fewer unnecessary tests ordered.
    • Avoidance of Medicare penalties for some readmissions.
    • Fewer medical errors.
    • Increased use of generic drugs.
    • Reduced malpractice.
  • Dissemination of expert knowledge:
    • Sharing of best evidence.
    • Education of all staff, students, and patients.
  • Management of complex clinical issues:
    • Use of clinical practice guidelines, smart forms, and order sets.
    • Interdisciplinary sharing of information.
    • Case management.
  • Monitoring clinical details:
    • Reminders for preventive services.
    • Tracking of diseases and referrals.
  • Improvement of population health:
    • Identification of high-cost/needs patients.
    • Mass customized messaging.
  • Management of administrative complexity:
    • Supports coding, authorization, referrals, and care management.
  • Support clinical research:
    • Ability to identify prospective research subjects.

Supporting Organizations

  • Institute of Medicine (IOM): promoted automated clinical information and CDS.
  • AMIA: developed 3 pillars of CDS in 2006—best available evidence, high adoption and effective use, and continuous improvement.
  • ONC: has funded research to promote excellent CDS and sharing possibilities.
  • AHRQ: also funded multiple CDS research projects and initiatives.
  • HL7: has a CDS working group and developed FHIR standards.
  • National Quality Forum (NQF): developed a CDS taxonomy.
  • Leapfrog: promoted both CPOE and CDS.
  • HIMSS: EMR Adoption Model rates EMRs from 0-7; full use of CDS qualifies as level 6.
  • CMS: Meaningful Use, Stages 1 and 2 include CDS measures.

CDS Methodology

  • Two phases of CDS: knowledge use and knowledge management.
  • Knowledge Use:
    • Triggers: an event, such as an order for a medication.
    • Input data: information within the EHR, such as patient allergies.
    • Interventions: CDS actions such as displayed alerts.
    • Action steps: overriding the alert or canceling an order for a drug to which the patient is allergic.
  • Knowledge Management:
    • Knowledge acquisition: acquire expert internal (EHR data) or external data (e.g., Apache scores) for CDS.
    • Knowledge representation: use expert information, integrate it with an inference engine, and communicate it to the end-user (e.g., an alert).

Knowledge Representation

  • Configuration: knowledge is represented by choices made by the institution.
  • Table-based: rules are stored in tables; if a current drug on a patient is in one row and an order for a second inappropriate drug is stored in the same row, an alert is triggered for the clinician.
  • Rules-based: knowledge base has IF-THEN statements; if the patient is allergic to sulfa and sulfa is ordered, then an alert is triggered. Earlier CDS programs, such as Mycin, were rule-based.
  • Bayesian networks: based on Bayes Theorem of conditional probabilities; predicts future (posterior) probability based on pre-test probability or prevalence. The formula is included below:
    • P(A|B) = \frac{P(B|A) * P(A)}{P(B)}
  • Knowledge-based CDS: based on known data.
  • Non-knowledge based CDS: based on data mining-related techniques.

Data Mining

  • Data mining (machine learning) algorithms must be developed and validated ahead of actual implementation.
  • Supervised learning: assumes the user knows the categories of data that exist, such as gender, diagnoses, age, etc.
    • Classification model: if the target (outcome or dependent variable) is categorical (nominal, such as lived or died).
    • Regression model: if the target is numerical (such as size of tumor, income, etc.).
    • Neural networks: configured like a human neuron; the model is trained until the desired target output is close to the desired target. Requires great expertise.
    • Logistic regression: used where the desired output/target is binary (cancer recurrence, no cancer recurrence). Multiple predictors are input, such as age, gender, family history, etc., and odds ratios are generated. The gold standard for much of predictive analytics.
    • Decision trees: can perform classification or regression and are the easiest to understand and visualize.
  • Unsupervised learning: data is analyzed without first knowing the classes of data to look for new patterns of interest.
    • Cluster analysis: one of the most common ways to analyze large data sets for undiscovered trends.
    • Association algorithms: look for relationships of interest.

Knowledge Maintenance

  • Constant updating of expert, evidence-based information is needed.
  • This task may fall to a CDS committee or technology vendor.

CDS Standards

  • Developers have struggled to share knowledge representation or modify rules locally.
  • Standards were developed to try to overcome these obstacles:
    • Arden syntax: represented by medical logic modules (MLMs) that encode decision information. The information can’t be shared because institution-specific coding resides within curly braces { } in the MLM (curly brace problem).
    • GELLO: can query EHRs for data to create decision criteria; part of HL7 v. 3.
    • GEM: permits clinical practice guidelines to be shared in an XML format, as an ASTM standard.
    • GLIF: enables sharable and computable guidelines.
    • CQL: draft HL7 standard to be used in XML format for electronic clinical quality measures (eCQMs).
    • Infobuttons: can be placed in workflow where decisions are made with recommendations.
  • Fast Healthcare Interoperability Resources (FHIR): developed by HL7; there is great hope that this standard will solve many interoperability issues.
    • It is a RESTful API (like Google uses) that uses either JSON or XML for data representation.
    • It is data-centric, not document-centric; a clinician could place an HTTP request to retrieve just a lab value from EHR B instead of a CCDA.
    • EHR can also request decision support from software on a CDS server.
    • Approximately 95 resources have been developed to handle the most common clinical data issues.

CDS Functionality

  • CDSSs can be classified in multiple ways:
    • Knowledge and non-knowledge-based systems.
    • Internal or external to the EHR.
    • Activation before, during, or after a patient encounter.
    • Activated automatically or on-demand.
    • Alerts can be interruptive or non-interruptive.
  • Examples:
    • Patient Safety: Medication alerts, critical lab alerts, ventilator support alerts, improved drug ordering for warfarin and glucose, infusion pump alerts.
    • Cost: Reminders to use generic drugs or formulary recommendations, fewer duplications, reminders about costs of drugs, lab tests and imaging studies, reduced medication errors, reduced malpractice claims, better utilization of blood products.

CDS Functionality

  • **Patient Care: ** * Embedded clinical practice guidelines, order sets, and clinical pathways.
    • Better chronic disease management.
    • Identify gaps in recommended care.
    • Immunization aids.
    • Diagnostic aids.
    • Sepsis alerts.
    • Antibiotic duration alerts.
    • Prognostic aids.
    • Patient reminders
  • Disseminating Expert Knowledge:
    • Use of infobuttons for clinician and patient education.
    • Provide evidence-based medicine with embedded clinical practice guidelines and order sets.
  • Managing Complex Clinical Issues:
    • Reminders for preventive care for chronic diseases
  • -Care Management
  • Managing Complex Administrative Issues:
    • Predictive modeling based on demographics, cost, and clinical parameters.

Ordering Facilitators

  • Order sets: EHR templated commercial or home-grown orders that are modified to follow national practice guidelines. For example, a patient with a suspected heart attack has orders that automatically include aspirin, oxygen, EKG, etc.
  • Therapeutic support: include commercial products such as Theradoc and calculators for a variety of medical conditions.
  • Smart forms: templated forms, generally used for specific conditions such as diabetes. They can include simple check-the-boxes with evidence-based recommendations.
  • Alerts and reminders: classic CDS output that reminds clinicians about drug allergies, drug-to-drug interactions, and preventive medicine reminders.

Relevant Information Displays

  • Infobuttons, hyperlinks, mouse-overs: common methods to connect to evidence-based information.
  • Diagnostic support: most diagnostic support is external and not integrated with the EHR; such as SimulConsult.
  • Dashboards: can also be patient, and not population level, so they can summarize a patient’s status and thereby summarize and inform the clinician about multiple patient aspects.

CDS Sharing

  • There is no single method for CDS knowledge that can be universally shared.
  • The approach has been to either use standards to share the knowledge or use CDS on a shared external server.
  • Socratic Grid and OpenCDS are open-source web services platforms that support CDS.
  • The FHIR standard appears to have the greatest chance for success.

CDS Implementation Steps

  • Project Initiation:
    • Ensure clinical and non-clinical leadership are onboard and have a shared vision.
    • Ensure CDS is synced with organizational goals, patient safety/quality measures, and meaningful use objectives.
    • Determine the business case/value of CDS for the organization.
    • Determine feasibility from a manpower and financial standpoint and acceptance by clinicians.
    • Ensure objectives are clear and attainable.
    • Identify key stakeholders and assess buy-in.
    • Understand that the CDS needs of specialists are different from primary care.
    • Assess readiness, EHR capability, and IT support.
    • Assess the clinical information systems (CISs) involved.
    • Assess knowledge management capabilities.
    • Assemble the CDS team: clinical leaders, CMIO, administrative and nursing leaders, managers, EHR vendor, and IT experts.
    • Identify clinical champions.
    • Develop CDS charter.
  • Project Planning:
    • Consider a SWOT analysis (strengths, weaknesses, opportunities, and threats).
    • Utilize standard planning tools such as Gantt charts and swim lanes.
    • Develop timeline.
    • Decide whether to build or buy CDS content.
    • CDS committee should select CDS interventions that fit their vision.
    • Be sure to follow the 5 Rights of CDS.
    • Map the different processes involved with CDS and be sure they integrate with the clinician's workflow.
    • Determine whether you will measure structure, processes, and/or outcomes.
    • Plan the intervention: triggers, knowledge base, inference engine, and communication means.
    • Educate staff and gain their input.
    • Design the CDS program for improvement over baseline performance in an important area for the organization. In other words, be sure you can measure outcomes and compare with baseline data.
    • Investigate the needed CDS standards required.
    • Follow the mandates of change management, e.g., John Kotter's Eight Step Model.
    • Communicate goals of CDS project to all affected.
  • Project Execution:
    • Provide adequate training and make CDS training part of EHR training.
    • Develop use cases.
    • Test and re-test the technology: unit, integration, and user acceptance testing.
    • Decide on incremental roll-out or "big bang".
    • Provide a mechanism for feedback in the CDS process, as well as formal support.
  • Project Monitoring and Control:
    • Use data from feedback, override logs, etc. to modify the system as needed.
    • Compare the alert and override rates with national statistics.
    • Measure percent of alerts that accomplished desired goals.
    • Communicate the benefits and challenges to the end-users as they arise.
    • Use tools such as the AHRQ Health IT Evaluation ToolKit.
    • Knowledge management maintenance; are guidelines unambiguous and up to date? Who will maintain the content?

CDS Challenges

  • General: exploding medical information that is complicated and evolving; tough to write rules.
  • Organizational support: CDS must be supported by leadership, IT, and clinical staff. Currently, only large healthcare organizations can create robust CDSSs.
  • Lack of a clear business case: evidence shows CDS helps improve processes, but it is unclear if it affects behavior and patient outcomes. Therefore, there may not be a strong business case to invest in CDSSs.
  • Unintended consequences: alert fatigue.
  • Medico-legal: adhering to or defying alerts has legal implications; product liability for EHR vendors.
  • Clinical: must fit clinician workflow and fit the 5 Rights.
  • Technical: complex CDS requires an expert IT team.
  • Lack of interoperability: must be solved for CDS to succeed.
  • Long-term CDS benefits: requires long-term commitment and proof of benefit to be durable.

Lessons Learned

  • Project Initiation:
    *Ensure the organization can support a new CDS (Healthcare organizations have competing initiatives. Even if CDS is intended to match meaningful use, it must be embraced by all and match organizational goals).
    *CDS cannot come from external mandate
  • Project Planning:
    • Customization of content and workflow is important (One size CDS does not fit all).
    • CDS must match the 5 Rights of CDS.
    • Make CDS as non-intrusive and interruptive as possible.
      *Customization is desirable but labor intensive and not available at smaller organizations.
      *Ideally, there should be recommendations for clinicians and patients (Specialists and primary care clinicians have different needs. Clinicians do not want to stop and speed is important. Table 24.1 CDS Taxonomy)
      *Interventions should include a reason for and speed interventions override intervention should make recommendation and not just assessment
      "Do CDS with users, not to them"
      *EHR data must be up to date for triggers to work correctly
  • Project Execution:
    *Feedback buttons in CDS work well
    *Include CDS training into EHR training
    *CDS must be tested for UACS and patient safety
    *User feedback is critical
  • Project Monitoring and Control:
    • Knowledge management is time-consuming.
    • Be sure intervention content is up to date.

Future Trends

  • The future of Meaningful Use is unclear, so there is no obvious CDS business case for clinicians, hospitals, and vendors.
  • If the FHIR standard makes interoperability easier, we may see new CDS innovations and improved adoption.

Conclusions

  • CDS could potentially assist with clinical decision-making in multiple areas.
  • While there is widespread support for CDS, there are a multitude of challenges.
  • CDS is primarily achieved by larger healthcare systems.
  • The evidence so far suggests that CDS improves patient processes and to a lesser degree clinical outcomes.