Comorbidities as predictors of inpatient deaths after hip fracture in Chinese elderly patients: analysis of hospital records

Abstract Background Hip fractures in elderly patients are common and are associated with substantial mortality and variations in outcome. We aimed to describe the type and prevalence of comorbidities in elderly patients (aged 55 years or older) with hip fracture, develop and evaluate multivariable risk-adjustment models for in-hospital mortality, verify if inpatient mortality associated with specific comorbidities is relatively higher in surgical patients compared with those not associated with specific comorbidities, and assess the clinical application of a preoperative risk scoring system for elderly patients with hip fracture. Methods From Jan 1, 2006, to Jan 1, 2010, we used hospitalisation summary reports from the 35 top-rank hospitals in Beijing to identify 14 577 patients aged 55 years or older with hip fracture. We used multivariable regression methods to develop a risk-adjustment model for inpatient mortality in surgical patients. From these models, we obtained a preoperative risk scoring system for elderly patients with hip fracture. Preoperative risk scoring was performed with an evaluation form according to sex, age, and comorbidities. Patients were divided into risk groups according to score: very low risk (score of 0–10), low risk (score of 11–20), moderate risk (score of 21–30), high risk (score of 31–40), and very high risk (score of 41 or more). Postoperative complications and death before discharge from hospital were recorded. Findings From Feb 1, 2014 to Feb 1, 2016, 286 elderly patients with hip fracture who matched the inclusion and exclusion criteria were recruited from Peking University People's hospital. All 286 patients were treated with surgery. The cohort included 175 patients with femoral neck fracture and 111 patients with intertrochanteric fracture. Other than older age and male sex, ten different comorbidities were independently associated with in-hospital mortality, particularly history of pulmonary embolism, respiratory failure, renal failure, and diabetes with complications. The preoperative risk scores of 286 patients ranged from 0 to 69 (mean 24·90 [SD 10·96]). 44 patients had postoperative complications and five patients died during hospitalisation. The frequency of complications was 4% (1/25) in the very low risk group, 4% (3/72) in the low risk group, 15% (17/116) in the moderate risk group, 24% (12/51) in the high risk group, and 50% (11/22) in the very high risk group. Interpretation Risk of inpatient mortality associated with specific comorbidities is relatively higher in surgical patients compared with those without specific comorbidities, particularly pulmonary embolism, respiratory failure, renal failure, and diabetes. The frequency of postoperative complications increased in line with preoperative risk scores in elderly patients with hip fracture, manifesting a positive correlation effect. The preoperative risk scoring system had excellent accuracy for predicting in-hospital mortality and postoperative complications, but further improvement and evaluation of clinical application in a multicentre study is still needed. Funding Chinese National Ministry of Science and Technology 973 Project Planning (No. 2014CB542206) and 863 Project Planning (SS2015AA020501), the National Natural Science Fund (31271284, 31171150, 81171146, 31471144, 30971526).