Phase of care CDS tool Example Pre-visit Referral recommendation Pre-visit referral used to expedite antibiotic Patient screening survey preparation for oncology patient with fever Assessment/Plan Pattern recognition Low-risk head injury rule formulation implemented into Knowledge delivery nursing and provider assessments Plan execution Order set Order set enables implementation of Error checking metered dose inhaler to lower admission rate for mild-moderate asthma Response to events Alerts Sequential alerts reduce patients missed for Enhanced data display treatment on sepsis pathway Discharge Education guides Video-based education integrated into Discharge order sets discharge process improves fever knowledge and satisfaction Post-visit Patient reminders Communication to communication Communication to primary care providers enhances providers communication of behavioral health screening results Adapted from Osheroff J, Teich J, Levick D, et al Improving Outcomes with Clinical Decision Support: An Implementer’s Guide 2nd ed HIMSS Publishing; 2012 Plan Execution The plan execution phase of the encounter is one of the most commonly used approaches to provide CDS through order sets, order questions and error-checking after order entry Reviews of the literature have found the strongest evidence of CDS benefit in improving processes in this area, although studies have limitations and have rarely assessed outcomes of care Although these forms of CDS can work well, they may be less effective if attempting to influence a plan that has already been made during the assessment phase, as discussed above One of the “Ten Commandments” of CDS described in a Bate’s classic publication is to “recognize that physicians will strongly resist stopping.” A better approach is often to attempt to change their direction further upstream in the workflow when possible Order sets are collections of orders grouped together for use in a clinical process, typically for a specific indication or patient population Order sets work particularly well in the ED setting given the predictable sequence of patient arrival, assessment, and treatment, which can allow them to be tailored to a chief complaint or diagnosis Benefits of order sets include the ability to increase efficiency by grouping orders together and preparing them for a specific use, which may allow orders to be prefilled or prechecked in ordering systems Order sets can reduce variation between providers by producing a consistent set of orders that have been planned in advance They may serve an educational role for trainees by laying out the expected plan for a specific condition This may be particularly effective when order sets are coupled with an overall algorithm, care plan, or pathway Potential hazards of order sets reflect the converse of the benefits Prefilled orders may not fit all scenarios and may be ordered inadvertently for an inappropriate clinical situation Trainees may become reliant on order sets and develop “automation bias” in which active learning may be inhibited by the expectation that a safety net is provided by the EHR Key principles for effective order set design include the following: Naming and Suggestion: Order sets need to be thoughtfully named to make them easy to find Prior literature has shown that typically only a fraction of available order sets is actually used by clinicians In the ED setting, where order sets are often used for suspected conditions, placing the disease entity first in the title may ease searching for the order set (i.e., Sepsis suspected), as well as including synonyms to allow them to be searched under multiple terms Some EHR systems may allow suggestion of order sets by chief complaint or other criteria, which can greatly enhance order set use Defaults: Careful attention should be paid to prechecking of orders in order sets For an order to be prechecked, it should be appropriate in virtually all scenarios for which the order set would be used If a choice should be made by a clinician, ideally the EHR will allow this to be preconfigured to require that choice through questions or layered panels of orders Recommendations to Not Order: Recommendations against routine use of treatments may be best accomplished by including them in the order set and providing specific indications for their appropriate use rather than omitting them At our institution, a quality improvement project following clinical guideline recommendations against use of albuterol for bronchiolitis found sustained reductions when this recommendation against use was incorporated in the order set (described as the “Do Not Order Set”) Testing and Review: Order sets should be tested prior to release with providers who were not involved in their development Many institutions have independent committees that review all order sets, alerts and other forms of CDS before their release to insure adherence to recommended principles An example of effective use of CDS is provided in a quality improvement project by Abaya et al to improve ED treatment of mild to moderate asthma exacerbations A respiratory chief complaint suggested the order set, which had a link to the clinical pathway that displayed the algorithm for the change in treatment Corticosteroids were prechecked to encourage early ordering but required the provider to stop to consider the correct agent and dose Based on a general assessment, the provider could choose one of severity subsections with a prechecked order for metered dose inhaler for mild or moderate patients to increase use of this approach The orders included an initial treatment and protocol orders for the respiratory therapist to reassess and stop if the patient was improved The combined impact of this CDS intervention was associated with an almost immediate change in practice accompanied by a sustained decrease in length of stay and admission rate for asthma Response to Events Another common CDS approach is the use of alerts that are triggered by data elsewhere in the system Most EHR systems have robust ability to trigger alerts by a variety of actions, including clinicians opening the chart, placing orders, performing documentation, receipt of interface messages from external systems, or execution of complex rules combining multiple of these data elements Unfortunately, the relative ease of designing alerts has led to their overuse Alert fatigue, the inattention of clinicians who are overburdened with alerts from EHRs, monitors and other systems, has been identified as a leading safety hazard in healthcare by The Joint Commission and other organizations Key points in alert design include the following: Five Rights of Alert Design: The design of an alert should pay careful attention to the underlying process being impacted Osheroff and colleagues have described “5 Rights” of an effective alert that can be systematically applied Table 3.2 presents an example of two EHR alerts designed by Balamuth et al to sequentially appear for nurses and physicians that successfully increased the use of a sepsis pathway and order set for patients with high risk for sepsis Information: An alert should clearly display to the user the information that caused it to be fired Ambiguity about the appropriateness of an alert can undermine trust from clinicians In the sepsis example, the nursing alert displayed the documented vital signs, elevated heart rate or low blood pressure, which caused the initial alert to fire EHR systems can often easily display a variety of data in alerts such as lab results (e.g., most recent serum creatinine level when alerting about a medication with renal toxicity) that may be of great salience when given appropriately Time: As discussed above, timing in the work process may be key to influencing and gaining acceptance from clinicians For the nursing sepsis alert, firing an alert as soon as an elevated heart rate was entered would interrupt the process of entering vital signs, which was