Effects of CPOE on provider cognitive workload: a randomized crossover trial
This is a review of Effects of CPOE on Provider Cognitive Workload: A Randomized Crossover Trial 
The authors evaluated the effects of systematically developed clinical decision supports to determine their effect on cognitive load, system usability, and patient safety.
The authors describe clinical decision support as “a process for enhancing health-related decision making by providing pertinent, evidence-based clinical knowledge at the time such knowledge is to be used”. An order set, a type of clinical decision support tool, contains a number of patient care interventions or activities such as medications, therapies, laboratory studies, imaging studies, and nursing orders. Order sets are created to simplify computerized physician order entry. At the time of this study, there were no published standards or best practices for developing order sets, although they are in use in many facilities using electronic health records (EHR). Seattle Children’s Hospital (SCH) developed order sets for postsurgical appendicitis based on current clinical guidelines using a multidisciplinary approach.
SCH is a 250-bed regional medical center with approximately 250 admissions per year for appendicitis. Study participants were volunteer surgeons, surgical fellows, and surgical residents. This study was performed on test patients using a randomized, crossover approach. Participants were given two clinical scenarios (perforated and nonperforated appendicitis) and asked to enter orders for postsurgical care on each patient. One trial required the participants to use ad hoc ordering and the other trial required them to use the standardized order set. The assignments were reversed (the crossover) so each participant was observed using order sets and ad hoc ordering. The participants were not aware of the purpose of the study.
Participants were videotaped while performing these trials. Observers counted mouse clicks and keystrokes and noted usage patterns, completion time, and amount of free text entered in association with the orders. Participants were asked to verbalize their thought processes. Immediately following the trials, participants completed surveys including the system usability scale (SUS) and the National Aeronautics and Space Administration’s Task Load Index (NASA-TLX). The SUS uses a 5-point Likert scale to subjectively assess usability. The NASA-TLX measures mental demands, physical demands, temporal demands, own performance, effort, and frustration. Finalized orders were evaluated for appropriateness, variability among participants, and conformance to evidence-based postsurgical guidelines.
When compared to ad hoc order sets, the systematically designed order sets were found to be easier to use, with reduced cognitive workload, less variation among surgeons, and greater adherence to guidelines. All of the providers preferred using the systematically designed order sets over the ad hoc order set approach. One unintended consequence discovered during this study was confusion about the degree of tailoring required once an order set had been selected; some participants did not realize the orders still needed to be adjusted for the patient’s weight. Initially, mouse click counts were not found to be lower when using the predesigned order sets, which seemed counterintuitive, but further evaluation showed that users were clicking the down buttons repeatedly instead of scrolling through the order sets. Therefore mouse clicks were not a good indicator of actual “work” but rather a reflection of navigation techniques. Once mouse-click counts were adjusted for scrolling, use of the predesigned order sets appeared to be more efficient.
The findings of this study suggest that systematically developed, predesigned order sets can improve patient care by aligning practices with the Institute of Medicine’s six tenants of quality care: improved safety, improved effectiveness (via conformance to evidence-based clinical guidelines), improved efficiency, and improved equity (treating all similar patients the same). Clinical work is reduced through decreased physical demand (compared to searching for guidelines and entering orders ad hoc) and decreased mental demand (by making guidelines available at the time of need). Reducing clinical work in this way increases provider satisfaction and increases adherence to guidelines.
The authors noted that positive outcomes with standardized order sets require the provider to have confidence in the accuracy and appropriateness of the order sets. Without trust, the provider will not utilize the standardized order sets or will spend additional time and energy double-checking and/or correcting them.
To gain efficiency, order sets also included alternative medications to make it simpler to substitute medications when first-line medications are contraindicated due to allergies or comorbidities, therefore, order sets included prechecked items as well as items that must be actively checked. To reduce the potential for harm, SCH established a policy that prechecked options within the order sets must apply to greater than 95% of patients and the remaining 5% must not be harmed by prechecked options.
The authors concluded that systemically developed order sets can improve quality of care and decrease clinician cognitive load, but order sets must be based on sound clinical guidelines and developed carefully using a multidisciplinary approach. The authors recommend that facilities establish best practices for the development of standardized order sets and institute policies for their use and periodic review.
Lessons I learned from this paper include the value of policies and procedures for developing standardized order sets (as opposed to allowing each physician to develop their own), the need for best practices and industry guidance for implementing order sets, and the value of creating a trustworthy process. Another interesting takeaway is how the authors tied the outcomes to the Institute of Medicine’s tenants of quality care: patient safety, effectiveness, efficiency, and equity. Lastly, this study showed that the number of mouse clicks does not necessarily correlate to the level of work.
- Avansino, J., & Leu, M. G. (2012). Effects of CPOE on provider cognitive workload: a randomized crossover trial. Pediatrics, 130(3), e547-e552. http://pediatrics.aappublications.org/content/130/3/e547.short