Hospitals increasingly rely on electronic medical records to embed standardised care pathways for acute conditions. An evaluation at a large quaternary-care hospital examined how a clinical decision support system (CDSS) care plan embedded in the electronic medical record was used during management of diabetic ketoacidosis (DKA) and how utilisation related to patient outcomes. Across hundreds of DKA episodes, the care plan was used in just over half of cases, with wide variation in how extensively orders were placed through the tool. Regression analyses linked greater utilisation to shorter length of stay and any use to lower in-hospital mortality, while readmissions showed no significant association. Interviews with clinicians provided context on adoption, training and workflow fit, highlighting opportunities to improve meaningful use.
Must Read: Addressing Alert Fatigue to Improve CDSS Acceptance
Usage Patterns and Setting
The evaluation drew on routinely collected data from a large adult quaternary-care hospital that has used an integrated electronic medical record since 2017, including standardised care plans within its CDSS. DKA was selected because it is clinically complex, time sensitive and susceptible to unwarranted clinical variation. Data covered all DKA episodes from January 2018 to June 2021, comprising 345 episodes among 235 unique patients and 48,757 linked order records. Two utilisation measures were analysed: a binary indicator of whether the CDSS-embedded care plan (Cerner PowerPlan) was used at all and a continuous measure representing the proportion of DKA-related orders placed via the care plan during each episode.
Of the 345 episodes, 151 involved at least one order via the CDSS and 194 did not. Among episodes with any use, clinicians placed on average about six orders through the tool and, overall, used the CDSS for roughly 8% of DKA-related orders, indicating considerable room to increase within-episode utilisation. Descriptive characteristics differed between groups: CDSS episodes involved younger patients with slightly lower comorbidity scores and showed lower in-hospital mortality but longer crude length of stay and higher crude 30-day readmission rates, patterns that motivated adjusted analyses.
Measured Impact on Outcomes
Adjusted regression models associated care plan utilisation with shorter hospital stay. Any use corresponded to an average 14.8% reduction in length of stay and the continuous measure suggested that each 1% increase in utilisation related to an additional 1.3% reduction, holding other factors constant. Mortality analyses indicated that any use of the care plan was linked to a 3.3% reduction in in-hospital mortality on average, whereas the proportional utilisation measure did not show a significant relationship with mortality after adjustment. No significant association emerged between either utilisation measure and 30-day readmission, which may reflect that readmissions outside the facility were not captured. These results were estimated with controls for patient characteristics, total DKA-related orders as a proxy for complexity, admitting service and ward and were supported by robustness checks using alternative specifications.
Taken together, the analyses suggest that engaging the care plan at all is associated with clinically relevant benefits for DKA, particularly reduced length of stay and lower in-hospital mortality, while deeper incremental use within an episode aligns most strongly with shorter stays. The absence of a detected effect on readmission does not preclude potential benefits beyond the measured setting, given the limitations in tracking re-presentations to other hospitals.
Barriers and Enablers of Meaningful Use
Semi-structured interviews with clinicians and informatics stakeholders provided context for the observed patterns. Participants generally viewed the care plan positively, citing value for standardising care, saving time, supporting adherence to protocols and preventing missed orders. Many described use of the initial phase as efficient for starting fluid resuscitation, insulin therapy and electrolyte replacement, particularly for junior clinicians. Yet later phases were used less consistently as some orders were charted outside the plan once initial steps were underway.
Limited awareness and lack of formal training were frequently cited barriers. New staff rotations, reliance on on-the-job teaching and perceived complexity of the interface reduced consistent uptake across shifts and teams. Senior clinical leadership played a notable role in promoting use, embedding expectations into local culture and workflows. While flexibility to tailor orders was valued to accommodate warranted clinical variation, adaptations such as favourited plans could perpetuate both good and suboptimal practices if not governed carefully. Interviewees also pointed to the need for clearer access to linked guidance within the plan and for design refinements that reduce visual complexity without diluting decision support.
A CDSS-embedded care plan for DKA was associated with shorter length of stay and lower in-hospital mortality when used, yet utilisation was inconsistent and concentrated in early care phases. The combination of outcome signals and clinician feedback points to practical levers for improving value: raise awareness, provide targeted training, reinforce expectations through leadership and streamline design to fit busy workflows. For organisations pursuing safer, more consistent acute care, aligning adoption efforts with workflow and education may be as consequential as the care plan content itself, helping convert embedded protocols into measurable benefits for patients with DKA.
Source: npj digital medicine
Image Credit: iStock
