From the Journals

Postop delirium management proposed as hospital performance measure


Key clinical point: Through predictive modeling, the study identified 20 risk factors markedly associated with delirium that can be used to identify high-risk patients.

Major finding: Among the 2,427 patients who experienced delirium, 35% had preoperative cognitive impairment, 30 % had a surrogate sign the consent form, and 32% experienced serious postoperative complications or death.

Study details: An analysis of 2,427 elderly patients at 30 hospitals through data from the ACS NSQIP Geriatric Surgery Pilot Project.

Disclosures: This project is funded in part by a grant from the John A. Hartford Foundation. The authors declare no conflict of interests.

Source: Berian JR et al. Ann Surg. 2017Jul 24. doi: 10.1097/SLA.0000000000002436



A study suggests that delirium in older adults is a postoperative variable that should be included in surgical quality registries, outcome measures, and assessment of hospital performance.

Lead author Julia R. Berian, MD, of the University of Chicago Medical Center and her colleagues wrote, “Postoperative delirium has been associated with mortality, morbidity, prolonged length of stay, and increased costs of care. Furthermore, postoperative delirium may be associated with long-term cognitive and functional decline. However, postoperative delirium has not been incorporated as an outcome measure into major surgical quality registries. Approximately one-third of hospitalized delirium is believed to be preventable, making postoperative delirium an ideal target for surgical quality improvement efforts,” Dr. Berian and her colleagues reported in the Annals of Surgery.

Dr. Julia R. Berian is a surgery resident at the University of Chicago

Dr. Julia R. Berian

The investigators analyzed medical records from elderly patients at 30 hospitals using data from the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) Geriatric Surgery Pilot Project for whom postoperative delirium status was ascertained as a binary outcome. The Geriatric Surgery Pilot Project collects and studies novel, geriatric-specific variables on patients aged 65 years and older to include in the NSQIP. The team’s study sample included 20,212 patients aged 65 years and older undergoing surgeries from Jan. 1, 2014, through Dec. 31, 2015. Investigators used hierarchical logistic regression models adjusted for case mix and patient risk factors to measure for hospital performance on managing postoperative delirium in elderly patients. Model performance was then assessed with Hosmer-Lemeshow test and C-statistics, and compared across surgical specialties.

The Geriatric Surgery Pilot data abstractors were instructed to assign postoperative delirium if the medical record words indicating an acute confusional stat such a mental status change, confusion, disorientation, agitation, delirium, and inappropriate behavior. Data were collected from the period 2 hours after surgery to exclude effects of the pharmacologic agents of anesthesia. Delirium status was ascertained as a binary outcome (Yes/No).

Postoperative delirium was observed in 2,427 patients for an average, unadjusted rate of 12.0%. Investigators identified 20 risk factors markedly associated with delirium. The strongest predictors included preoperative cognitive impairment, preoperative use of mobility aid, surrogate consent form, ASA class 4 or greater, age 80 years and older, preoperative sepsis, and fall history within 1 year. Patients with delirium generally were older than patients without delirium were and accounted for a greater proportion of emergency cases. Postoperative hospital length of stay was about 4 days longer on average for patients with delirium, compared with those without delirium.

By specialty, the highest rates of postoperative delirium occurred following cardiothoracic (13.7%), orthopedic (13.0%), and general surgeries (13.0%). Study authors found varied associated risk for postoperative delirium within each surgical specialty. For example, in general surgery, the risk for postoperative delirium with partial mastectomy was low, compared with a mid-level risk in the repair of a recurrent, incarcerated, or strangulated inguinal hernia and a high-level risk in Whipple operations.

The model developed to measure delirium management success in 30 hospitals found that adjusted delirium rates ranged from 3.2% to 27.5%, with eight poor- and five excellent-performing outliers. Authors noted that their model demonstrated good calibration and discrimination. Examination of changes in the Bayesian Information Criteria indicates that as few as 10-12 variables may suffice in building a parsimonious model with “an excellent fit.”

Next Article:

   Comments ()