The use of an electronic clinical decision support tool called “ePNa” reduced severity-adjusted, 30-day, all-cause mortality by 38% across 16 community hospitals in Utah compared with predeployment levels, a 3-year, pragmatic, cluster-controlled study shows.
“We designed the ePNa specifically to require minimal input from the clinician so everything it does is already in the electronic medical record (EMR),” Nathan Dean, MD, University of Utah School of Medicine, Salt Lake City, told Medscape Medical News.
“So it’s actually putting the guideline recommendations into effect for physicians so that they can make better decisions by having all this information — it’s a comprehensive best practice kind of tool where best practices are likely to make the biggest difference for patients with a high severity of illness,” he added.
The study was published online in the American Journal of Respiratory and Critical Care Medicine.
The ePNa makes use of pneumonia guidelines of 2007 and 2019 from the American Thoracic Society/Infectious Disease Society of America. The system was deployed into six geographic clusters of 16 Intermountain hospital emergency departments (EDs) at 2-month intervals between December 2017 and November 2018. Simultaneous deployment was impractical, as implementation of the tool takes education, monitoring, and feedback that can be facilitated by focusing on only a few hospitals at a time.
The decision support tool gathers key patient indicators including age, fever, oxygen saturation, vital signs, and laboratory and chest imaging results to offer recommendations on care, including appropriate antibiotic therapy, microbiology studies, and whether a given patient should be sent to the intensive care unit (ICU), admitted to hospital, or may safely be discharged home.
Investigators analyzed a total of 6848 patients, of whom 4536 were managed for pneumonia before the ePNa was deployed and 2312 after deployment.
The median age of patients was 67 years (interquartile range, 50 – 79 years). Roughly half were female and almost all were White. “Observed 30-day all-cause mortality including both outpatients and inpatients was 8.6% before deployment versus 4.8% after deployment of ePNa,” Dean and colleagues report.
Adjusted for severity of illness, the odds ratio (OR) for lower mortality post-ePNa launch was 0.62 (95% confidence interval [CI], 0.49 – 0.79; P < .0010) “and lower morality was consistent across hospital clusters.”
Compared with patients who were discharged home, reductions in mortality were greatest in patients who were directly admitted to ICUs from the ED (OR, 0.32; 95% CI, 0.14 – 0.77; P = .01). The OR for patients admitted to the medical floor was 0.53 (95%CI, 0.25 – 1.1; P = .09), which did not reach statistical significance.
Dean explained that the reductions in mortality were seen among those with the most severe illness, in whom best practices would benefit the most. In contrast, patients who are sent home on an antibiotic are at low risk for mortality while patients admitted to the medical floor may well have another, more lethal illness from which they end up dying, rather than simple pneumonia.
“For me, this was a clear demonstration that these best practices made the biggest difference in patients who were sick and who did not have any underlying disease that was going to kill them anyway,” he emphasized. On the other hand, both 30-day mortality and 7-day secondary hospital admission were higher among patients the tool recommended for hospital ward admission but who were discharged home from the ED.
“This was an unexpected finding,” Dean observed. However, as he explained, the authors reviewed 25% of randomly selected patients who fell into this subgroup and discovered that the ePNa tool was used in only about 20% of patients — “so doctors did not use the tool in the majority of this group.”
In addition, some of these patients declined hospital admission, so the doctors may have recommended that they be admitted but the patients said no. “The hypothesis here is that if they had been admitted to the hospital, they may have had a lower mortality risk,” Dean said.
Another noticeable change following the introduction of the ePNa tool was that guideline-concordant antibiotic prescribing increased in the 8 hours after patients presented to the ED, from 79.5% prior to the tool’s launch to 87.9%, again after adjusting for pneumonia severity (P < .001). Use of broad-spectrum antibiotics was not significantly different between the two treatment intervals, but administration of antibiotics active against methicillin-resistant Staphylococcus aureus dropped significantly between the two treatment intervals (P < .001). And the mean time from admission to the ED to the first antibiotic taken was slightly faster, improving from 159.4 minutes (95% CI, 156.9 – 161.9 minutes) prior to the ePNa launch to 150.9 minutes (95% CI, 144.1 – 157.8) postdeployment (P < .001).
“Overall outpatient disposition for treatment of pneumonia from the emergency department increased from 29.2% before ePNa to 46.9% [post ePNA],” the authors note, while a similar increase was observed in patients for whom ePNA recommended outpatient care — from 49.2% pre-ePNA to 66.6% after ePNA.
Both hospital ward admission and admission to the ICU decreased after ePNa had been introduced. Despite a significant increase in the percentage of patients being discharged home, neither 7-day secondary hospital admission nor severity-adjusted, 30-day mortality were significantly different before versus after the introduction of ePNa, the authors stress.
A limitation of the study was that the trial was confined to a single healthcare system in one region of the United States with a patient population that may differ from that in other regions.
Reason for Its Success
Asked to comment on the findings, Adam Balls, MD, emergency department chair, Intermountain Medical Center, Murray, Utah, suggested that the reason the ePNa tool has been so successful at improving care for pneumonia patients is that it puts the guidelines directly into the hands of individual providers and tells them what’s going on. (Balls was not involved in the study.) “The tool allows us to take into consideration various clinical features — a patient’s oxygen requirements and whether or not they had prior complicated pneumonias that required additional antibiotics, for example — and then it makes the best determination for not only the disposition for that patient but antibiotic treatment as well,” he told Medscape Medical News.
This then allows physicians to either appropriately discharge less severely ill patients and admit those who are more ill — “and in general, just do a better job of treating pneumonia with this tool,” Balls said. He himself uses the decision support tool when attending to his own patients with pneumonia, as he feels that the tool really does make his care of these patients better. “There is a disparity around how we treat pneumonia in the US,” Balls acknowledged.
“Clinicians sometimes have a bias or a preference for certain antibiotics and we may not be appropriately treating these patients with broad-spectrum antibiotics or are perhaps using antibiotics that are not as effective based on an individual patient scenario so this is definitely a user-friendly tool that hopefully can be deployed throughout other health care systems to improve the treatment of pneumonia overall,” Balls emphasized.
Am J Respir Crit Care Med. Published online March 9, 2022. Abstract
Dean and Balls have disclosed no relevant financial relationships.