CORONADO, CALIF. – Variation in readmission risk across hospitals following certain surgical procedures is more attributable to hospital factors than to patient characteristics, results from a large analysis demonstrated.
Such is the impact of the care delivery macro environment (CDM), which Sarah A. Brownlee and coauthors defined as a series of complex interactions between patient characteristics and imposed hospital attributes than can impact patient outcomes postoperatively.
The purpose of the current study was to determine the relative contribution of various aspects of the CDM to 1-year readmission risk after surgery. Working with colleagues Anai Kothari, MD, and Paul Kuo MD, in the One:MAP Section of Clinical informatics and Analytics in the department of surgery at Loyola University Medical Center, Ms. Brownlee analyzed the Healthcare Cost and Utilization Project State Inpatient Databases from Florida, New York, and Washington between 2009 and 2013, which were linked to the American Hospital Association Annual Survey from that same time period.
The researchers used smoothed hazard estimates to determine all-cause readmission in the year after surgery, and multilevel survival models with shared frailty to determine the relative impact of hospital versus patient characteristics on the heterogeneity of readmission risk between hospitals. They limited the analysis to patients aged 18 years and older who underwent one the following procedures: abdominal aortic aneurysm repair, pancreatectomy, colectomy, coronary artery bypass graft, and total hip arthroplasty.
Ms. Brownlee reported results from 502,157 patients who underwent surgical procedures at 347 hospitals. The 1-year readmission rate was 23.5%, and ranged from 12% to 36% across procedures. After controlling for procedure, the researchers observed a 7.9% variation in readmission risk between hospitals. Staffing accounted for 9.8% of variance, followed by hospital structural characteristics such as teaching status and clinical programs (7.5%), patient ZIP code (3.8%), hospital perioperative resources such as inpatient rehab (2.9%), hospital volume (2.8%), and patient clinical characteristics (2.1%). The following hospital characteristics were significantly associated with a lower risk of 1-year readmission: high physician/bed ratio (hazard ratio 0.85; P = .00017); transplant status (HR 0.87; P = .022); high-income ZIP code (HR 0.89; P less than .001); high nurse bed/bed ratio (HR 0.90; P = .047), and cancer center designation (HR 0.93; P = .021).
“Compared to patient clinical characteristics, hospital factors such as staffing ratios, perioperative resources, and structural elements account for more variation in postoperative outcomes,” Ms. Brownlee concluded. “However, it’s important to note that in the present study, over 70% of variation in readmission rates is not explained by the covariates that we analyzed. It’s possible that there are other factors we need to consider. That’s where the direction of this research is going. Much of the variation in readmission risk across hospitals cannot be characterized with currently utilized administrative data.”
The National Institutes of Health provided funding for the study. Ms. Brownlee reported having no financial disclosures.