• Print

SAFER Guides

SAFER: Safety Assurance Factors for EHR ResilienceThe Office of the National Coordinator for Health Information Technology
graphic element

Computerized Provider Order Entry with Decision Support

The Computerized Provider Order Entry with Decision Support SAFER Guide identifies recommended safety practices associated with Computerized Provider Order Entry (CPOE) and Clinical Decision Support (CDS). Completing this self-assessment in collaboration with a multi-disciplinary team will help an organization optimize the safety and safe use of CPOE with CDS in the EHR. The implementation and use of CPOE with CDS is complex and fragile, requiring careful planning, implementation, and maintenance to function properly. In the EHR-enabled healthcare environment, providers rely on technology to support and manage the complex processes related to CPOE with decision support, and this reliance creates potential safety risks that can be minimized by the adoption of the recommended practices in this guide.

The use of CPOE with decision support can improve medication safety as well as ensure that providers who electronically order diagnostic tests and consultations remain in the communication loop.1-8 However, certain CPOE-related practices can create safety risks.9-28 For example, partial adoption of CPOE, or a lack of CPOE monitoring (e.g., incomplete data entry or excessive use of free text), can create hazardous conditions.

CDS, whether stand-alone or integrated within an EHR, is designed to aid the clinical decision-making process at the point of care. The current scope of CDS focuses primarily on drugs, laboratory testing, radiology procedures, and clinical reference literature.29 Substantial evidence suggests that well-designed decision support not only enhances the quality of care, but directly improves patient safety by decreasing common errors and preventing omissions or missed opportunities that result in patient harm.3,30-33 In spite of this, many EHRs do not have robust or reliable decision support features, and poorly implemented IT systems have been shown to introduce errors that adversely affect care.9,13,18,23,34-37

Completing the self-assessment in the Computerized Provider Order Entry with Decision Support SAFER Guide requires the engagement of people both within and outside of the organization (such as EHR technology developers and diagnostic services providers). Because this guide is designed to help organizations prioritize EHR-related safety concerns, clinician leadership in the organization should be engaged in assessing whether and how any particular recommended practice affects the organization’s ability to deliver safe, high quality care. Collaboration between clinicians and staff members while completing the self-assessment in this guide will enable an accurate snapshot of the organization’s CPOE and CDS status (in terms of safety), and even more importantly, should lead to a consensus about the organization’s future path to optimize EHR-related safety and quality: setting priorities among the recommended practices not yet addressed, ensuring a plan is in place to maintain recommended practices already in place, dedicating the required resources to make necessary improvements, and working together to mitigate the CPOE-related safety risks introduced by the EHR.

Download the pdf Guide to access the checklist of recommended practices for self assessment and a supporting worksheet to identify action steps to achieve those recommended practices.

1

Coded allergen and reaction information (or No Known Allergies [NKA]) are entered and updated in the EHR prior to any order entry.39

More About This Practice
Rationale:

One of the main purposes of CDS is automated drug/allergy checking, which requires coded entry of allergies in the EHR.

Examples:
  • Users are reminded to enter patients' allergies or "no known allergies" before entering any medication orders.
  • A standard, controlled vocabulary of allergens and reactions (e.g., SNOMED-CT) is available and used.
  • There is a defined hierarchy of authority to edit or remove allergy-related information from a patient's EHR.
  • The EHR system permits entry of medication intolerances, distinguished from true allergies.
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
2

Evidence-based order sets are available in the EHR for common tasks/conditions and are updated regularly.38

More About This Practice
Rationale:

Order sets minimize errors of omission through standardization. Requiring clinicians to enter each of the individual orders for routine clinical practices increases the risk of overlooking one or more items.

Examples:
  • Order sets for medications, diagnostic tests, and procedures are developed on the basis of Institute For Safe Medication Practices (ISMP) guidelines.40
  • Order sets exist for the 10 most common clinical conditions (e.g., management of chest pain), procedures (e.g., insulin administration and monitoring), and clinical services (e.g., admission to labor & delivery).41
  • Clinical content is developed or modified based on evidence from authoritative sources, such as those in the AHRQ CDS Initiative or specialists within the organization.
  • EHR developer-provided clinical content is based on authoritative sources and is updated whenever those sources are updated.
  • Order sets for medications include complete pre-written medication orders ("order sentences") that include dose, dose form when necessary, route of administration, frequency, and a PRN flag and indication, if appropriate.39
  • Pre-written medication orders use doses that are weight-based, when appropriate.
  • Personalized order sets are not used. If an institution permits them, there is an annual review process, (e.g., clinical quality committee or medical director approval).
  • Medications requiring complex dosing guidelines (e.g., insulin sliding scale) are standardized and available electronically.
  • The CPOE list of orderable items (i.e., medication dictionary or orderable catalog) includes all formulary medications.
  • The CPOE list of orderable items includes acceptable, non-formulary medications, which are clearly marked, that users can order for out-of-formulary fulfillment.
  • Prescribing systems for children use weight-based dosing recommendations, age-appropriate dosing calculators and dose-range checking, and pediatric-specific drug-drug interaction alerts.
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
  • Pharmacy
3

User-entered orderable items are matched to (or can be looked up from) a list of standard terms.42

More About This Practice
Rationale:

CDS is important to patient safety. CDS can be supported by orders of standardized items, but not on free-text orders.

Examples:
  • Users can look up all orderable items (e.g., medications, laboratory, and radiology tests) and pick terms from lists instead of entering free-text. This should support various word orders (e.g., "abdominal ultrasound" or "ultrasound, abdominal"), various names (e.g., generic, brand, or synonym), and should be able to be browsed alphabetically.43
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
  • Pharmacy
4

The EHR can facilitate both cancellation and acknowledgment of receipt of orders for laboratory, radiology, and pharmacy.

More About This Practice
Rationale:

Communication errors, especially related to medication orders and diagnostic services, are frequent occurrences. Order tracking can reduce these errors.

Examples:
  • The user can look up whether the lab has received the specimen for testing or not.
  • When medication orders are canceled, information is received and acted on appropriately by the responsible pharmacy.
  • The 2-way interfaces that facilitate order tracking are tested pre- and post-go-live.
Suggested Sources of Input:
  • EHR developer
  • Health IT support staff
  • Diagnostic services
  • Pharmacy
5

CDS alerts are displayed in the relevant clinical context.44-49

More About This Practice
Rationale:

CDS to improve diagnostic or therapeutic decision-making should be accessible in real time at the point of care, otherwise, the advice generated may be useless or underutilized.50 Risks include information overload and clinician dissatisfaction.3,31,32

Examples:
  • A process is in place to identify and remove alerts that do not make sense in the particular clinical context. In some cases the process may require communication with the EHR developer.
  • Ambulatory alerts for cancer screening protocols should not be presented in the inpatient setting.51,52
  • Alerts for diabetic foot screening should not be presented for patients with bilateral lower extremity amputations.
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
  • Health IT support staff
6

CDS incorporates current "best practices" and guidelines from authoritative sources, such as national organizations and medical specialty professional associations.53

More About This Practice
Rationale:

Out of date or incorrect knowledge provided by the CDS system may be harmful.3,31,32

Examples:
  • For organizations that rely on EHR developer-provided CDS, a process is in place to ensure that CDS is based on authoritative sources and is regularly updated.
  • The expertise supporting CDS is demonstrated to EHR users before adoption.
  • Examples of authoritative sources include AHRQ's CDS Initiative and professional associations.
  • Colon cancer screening reminders follow U.S. Preventive Services Task Force guidelines.54
  • Vaccination reminders use the latest recommendations from the Advisory Committee on Immunization Practices.55
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
  • Health IT support staff
7

Clinicians are trained and tested on CPOE operations before being issued login credentials.

More About This Practice
Rationale:

CPOE is a complex tool. In order to maximize its safe and effective use, clinicians must be trained rigorously and should not be expected to "learn the basics on the job."

Examples:
  • Incentives such as continuing education (CME or CEU) credits are awarded for clinicians getting trained on CPOE.
  • Clinicians are required to demonstrate basic CPOE skills before getting their login credentials.56
  • Organizations evaluate whether specialized CPOE training should be required in high-risk areas.
  • Training is reinforced periodically, particularly with system changes/upgrades.
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
  • Health IT support staff
  • Pharmacy
8

Clinicians are engaged in implementing, reviewing, and updating CDS.53,57-61

More About This Practice
Rationale:

Failure to include clinicians in decisions that affect their clinical work environment, their decision-making capabilities, or how their decisions are communicated and recorded significantly increases the risk of hazardous events. CDS systems can be optimized through monitoring of use, overrides, and clinician satisfaction.

Examples:
  • Clinicians are involved in making (and keeping) the CDS content consistent with updated guidelines and algorithms. There is a process (that involves clinicians) to manage, evaluate, and prioritize CDS updates.53,60-63
  • Clinician-provided feedback is reviewed and used for refinement and maintenance of CDS and the relevant clinical content.53,59-61,63
  • Clinician overrides (i.e., decisions not to follow a computer-generated suggestion) for high-priority CDS elements are logged and available for review and reporting.64-66
  • For EHR developer-provided or controlled CDS, a process is in place to communicate about the need for CDS improvements with the developer.
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • Diagnostic services
  • Pharmacy
9

CPOE is used for ordering all medications, diagnostic tests, and procedures for which CPOE is available.38

More About This Practice
Rationale:

While full use of CPOE with advanced clinical decision support has been shown to reduce errors,50 partial use of CPOE can introduce errors.

Examples:
  • Except in unusual situations, providers are required to enter
    their orders into the CPOE system.
  • Exceptions (e.g., emergency orders in resuscitation
    situations) are clearly defined, and processes are in place
    (and followed) for their proper documentation in the EHR.
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
  • Health IT support staff
  • Diagnostic services
  • Pharmacy
10

There is minimal use of free-text order entry. Orders are entered and stored in standardized, coded form.38,67

More About This Practice
Rationale:

Free-text data can introduce errors if it is inconsistent with structured data or is not used or communicated properly. Free-text orders cannot be effectively supported with CDS.

Examples:
  • Organizational policy addresses safety precautions to be undertaken when free-text ordering is allowed.
  • When medications are entered using standardized, coded terms, corresponding narrative text is minimized. Processes are in place to ensure timely use and review of any narrative text.
  • When medications must be ordered using free text, as constrained by organizational policy, a pharmacist reviews the order to identify and address any drug-drug or drug-allergy interactions.
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
  • Health IT support staff
11

Order entry information is electronically communicated, such as through the computer or mobile messaging, to the people responsible for carrying out the order.68

More About This Practice
Rationale:

To have effective CPOE, orders must be electronically communicated. An automated process minimizes lapses in communication.

Examples:
  • Nurses are notified via the EHR when new results or orders are entered into the system for one of their patients (e.g., when they login to the system an alert tells them that new orders are available, or they are sent an informative page or text message).69
  • Orders that are not acknowledged by the individual responsible for carrying them out within appropriately defined time periods are automatically escalated to a supervisor.70
  • Workflow is evaluated to ensure that all electronic orders go to the intended recipient and that person documents their actions in the EHR.
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
  • Health IT support staff
12

Interruptive alerts, such as pop-ups at the time of ordering, are used with discretion and only for high-risk, high-priority conditions.44-49,60

More About This Practice
Rationale:

Excessive use of interruptive alerts creates clinician dissatisfaction and reduces their effectiveness, causing clinicians to miss important alerts.29

Examples:
  • For lower priority conditions, passive alerts that do not force an interruption of the workflow are available.47
  • High risk, high priority conditions that justify interruptive alerts are identified by clinicians and are subject to review.
  • Interruptive alerts at the point-of-care are used only after considering other available options.71
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
  • Health IT support staff
13

Drug-allergy interaction checking occurs during the entry of new medication orders and new allergies.50,67

More About This Practice
Rationale:

Interaction checking minimizes the risk of adverse drug events related to allergies.50,67

Examples:
  • Checking occurs when an ACE inhibitor is prescribed, to ensure that a patient with a history of ACE inhibitor-induced angioedema is protected.
  • Allergy checking occurs (for all current medications) whenever a new allergy is entered into the system.
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
14

Duplicate order checking occurs for high-risk medication, diagnostic tests, and procedure orders (excluding as needed "PRN" medications).50,67

More About This Practice
Rationale:

Duplicate order checking reduces the risk of inadvertent drug overdoses and unnecessary tests and procedures.50,67

Examples:
  • Therapeutic duplication checking occurs before new medication orders are submitted (e.g., two orders for the same or two different beta-blockers are placed).
  • Duplicate checking occurs before diagnostic tests or procedures are ordered.72
  • Duplicate checking does not include PRN (i.e., as needed) medication orders.
  • PRN orders should not include "overlapping" criteria (e.g., for pain 1-3 give aspirin AND for pain 2-4 give Vicodin).
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
15

Drug-condition checking occurs for important interactions between drugs and selected conditions.50

More About This Practice
Rationale:

Electronic drug-condition checking reduces the risk of preventable adverse drug events related to specific conditions.

Examples:
  • Drug-condition interaction checking occurs when new medications are ordered or new conditions are identified (e.g., Accutane or tetracycline prescribed for a pregnant woman).
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
16

Drug-patient age checking occurs for important age-related medication issues.13,92

More About This Practice
Rationale:

Drug-patient age checking reduces the risk of preventable age-related adverse drug events.

Examples:
  • Drug-patient age interaction checking occurs when new medication orders are submitted for dispensing (e.g., medications contraindicated in the elderly).
  • Changes in frequency, dose, or substitutions are suggested for more age-appropriate strategies.
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
17

Dose range checking (such as maximum single dose or daily dose) occurs before medication orders are submitted for dispensing.50,73

More About This Practice
Rationale:

Dose range checking reduces the risk of medication overdose.

Examples:
  • Renal dose adjustment suggestions along with information on the patient's renal status are clearly displayed prospectively for relevant medications.
  • Patient context (age, renal function) dynamically changes the defaults prospectively.
  • Maximum single dose and maximum daily dose are independently checked.
  • Dose limits are age and body size appropriate.
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
18

A process is in place to review interactions so that only the most significant interaction-related alerts, as determined by the organization, are presented to clinicians.46,47

More About This Practice
Rationale:

Tiered alerting by severity (significance) is associated with higher compliance rates for drug-drug interaction alerts.

Examples:
  • Less significant alerts are presented as information only, rather than as interruptive alerts.46
  • Alerts are modified in a dynamic fashion based on feedback from the users and monitoring of user behavior.
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
19

Clinicians are required to re-enter their password, or a unique PIN, to "sign" (authenticate) an order.

More About This Practice
Rationale:

Explicit order authentication reduces the risk of inadvertently entering orders under the wrong identity when someone else is logged in. It gives users an additional opportunity to confirm that the orders they entered are correct, and prevents them from inadvertently signing orders they did not intend to sign.

Examples:
  • An explicit authentication process occurs in addition to their original login for access to the EHR.
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
20

Corollary (or consequent) orders are automatically suggested when appropriate and the orders are linked together, so that changes are reflected when the original order is rescheduled, renewed, or discontinued.74

More About This Practice
Rationale:

Automatically suggested linked orders reduce order inconsistencies by managing closely associated orders in tandem.

Examples:
  • Examples include: Prothrombin time monitoring when warfarin is prescribed, or drug level measurements with Vancomycin or aminoglycoside orders.74
  • Corollary orders are deleted whenever the main order is deleted (e.g., if colonoscopy is cancelled, the bowel prep is also cancelled).
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
21

Users can access authoritative clinical reference materials directly from the EHR, including organization-specific information when available.42,53,59,60,62,75

More About This Practice
Rationale:

Ready access to information can reduce the risk of errors. CDS to improve diagnostic or therapeutic decision-making should be accessible in real time at the point of care; otherwise, the advice generated may be useless or underutilized.50

Examples:
  • Medication monographs (such as Micromedex), dosing calculators, diagnostic guides, laboratory reference materials, image atlases, anatomical diagrams, patient education materials, and disease-specific treatment guidelines are directly accessible from the order entry screen or module.76
Suggested Sources of Input:
  • EHR developer
  • Health IT support staff
22

CPOE and CDS functionality are tested to ensure proper operation before go-live and with test patients in the production system before clinical use.

More About This Practice
Rationale:

Appropriate testing reduces the risk of errors associated with inappropriate CDS or CPOE system behavior.

Examples:
  • A CPOE evaluation tool, such as the Leapfrog Group's CPOE "flight simulator" for hospitals (also known as the Texas Medical Institute of Technology's EHR-CPOE flight simulator), is used to evaluate the safety and effectiveness of CPOE and CDS functionality.77-79
  • CDS interventions are evaluated to ensure correct firing of alerts and reminders.80
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
  • Health IT support staff
23

Questions presented to the user by CPOE or CDS are unambiguous.50,81

More About This Practice
Rationale:

Misunderstanding queries posed by the system can lead to risks of errors and adverse events.82

Examples:
  • There are policies and procedures to evaluate the clarity of questions posed to users.
  • Questions should be kept simple and focused. For example, "Is IV contrast contraindicated?" may be confusing. It might be better to ask: Is IV contrast safe to administer? Yes, safe. No, not safe.
  • Avoid negatively and poorly worded questions such as "Do you want to cancel this alert? Yes, No, Cancel."
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
  • Health IT support staff
24

CPOE and CDS implementation and use are supported by usability testing based on best practices from human factors engineering.83,93

More About This Practice
Rationale:

Risks of untested usability include decreased clinician efficiency and clinician dissatisfaction, as well as errors and adverse events due to unintended consequences of CDS use.

Examples:
  • Major CDS and CPOE changes/interventions are tested with representative end users.83
  • Clinician-reported hazards associated with CPOE and CDS due to poor usability are regularly communicated to someone in a position to make improvements. Follow-up is monitored.
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
  • Health IT support staff
  • Diagnostic services
  • Pharmacy
25

Critical patient information is visible during the order-entry process.84

More About This Practice
Rationale:

Ensuring that critical data are visible in the EHR minimizes errors related to misidentification or failing to account for common clinical issues.

Examples:
  • Pertinent clinical information (age, weight, allergies, pregnancy status, creatinine clearance/GFR) as well as identifying patient information is displayed on or behind the ordering screen with no scrolling required to view all the pertinent clinical data.84
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
26

The clinician is informed during the ordering process when additional steps are needed to complete the order being requested.

More About This Practice
Rationale:

Clinicians may not be aware that an order will not be completed without additional steps, leading to delays in performing the order.

Examples:
  • Clinicians are informed when non-formulary medications require additional pre-approval.
  • Clinicians are informed when "send out" tests require special forms or procedures.
  • The mode of informing clinicians of incomplete orders could include passive notifications, such as an informative icon.
Suggested Sources of Input:
  • EHR developer
  • Diagnostic services
  • Pharmacy
27

Use of abbreviations and acronyms is minimized and standardized.85-87

More About This Practice
Rationale:

Acronyms and abbreviations are a source of errors in both paper and electronic records. Minimizing and standardizing use of acronyms and abbreviations reduces the risk of errors related to misunderstanding.

Examples:
  • Organizational policies on the use of abbreviations and acronyms incorporate, and are consistent with, their use in EHRs.
  • Use of abbreviations and acronyms is consistent with industry best practices.
  • Abbreviations such as qd or qid are avoided.
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
28

Additional safeguards, such as double check by a second specialist, are implemented in the EHR before high-risk medications are prescribed.

More About This Practice
Rationale:

Medication errors are the most common type of error that reach patients and cause harm. For high-risk medications, additional safeguards are justified to reduce the likelihood of harm.

Examples:
  • A clinician- or specialist-driven process is in place to identify high-risk medications that justify additional safeguards and integrate those safeguards into the EHR.
  • Chemotherapy agents require special authorization and are displayed in a visually distinct way (e.g., different color, italics, etc.).
  • TALLman lettering is used to reduce CPOE errors from orthographically similar medication names (i.e., look-alike or sound-alike medication names; acetaZOLAMIDE and acetoHEXAMIDE).88-90
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
  • Pharmacy
29

Key metrics related to CPOE and CDS (e.g., override rates) are defined, monitored, and acted upon to optimize safety and use.38,91

More About This Practice
Rationale:

Well-designed and correctly used CPOE and CDS can reduce the most common errors that harm patients. Monitoring and oversight of the performance and clinician use of CPOE and CDS functionality allows optimization of a powerful driver of improved patient safety in an EHR-enabled healthcare system.

Examples:
  • Key CPOE and CDS safety indicators, such as the
    following, are monitored and reported to leadership
    on a periodic basis:
    • Rates of preventable ADEs.
    • CPOE use rate.
    • Frequency (volume) of orders that generate an alert.
    • Override rate (% of alerts that are overridden)
      in comparison to alert volume.
    • Median turnaround time for STAT laboratory
      or radiology results.
    • Percent of all orders requiring modification by someone
      other than the ordering provider.
    • Alerts with the highest percent of overrides are evaluated
      on at least a quarterly basis for effectiveness and turned off
      if deemed unacceptable.
    • Usage of evidence-based order sets is monitored.
    • Clinician satisfaction with CDS alert functionality.
    • Results of any CPOE evaluation tool.
Suggested Sources of Input:
  • Clinicians, support staff, and/or clinical administration
  • EHR developer
  • Health IT support staff
{@Thursday, January 30, 2014@}
{@Wednesday, May 7, 2014@}