An ensemble machine learning approach to predict postoperative mortality in older patients undergoing emergency surgery

BACKGROUND

Prediction of preoperative frailty risk in the emergency setting is a challenging issue because preoperative evaluation cannot be done sufficiently. In a previous study, the preoperative frailty risk prediction model used only diagnostic and operation codes for emergency surgery and found poor predictive performance. This study developed a preoperative frailty prediction model using machine learning techniques that can be used in various clinical settings with improved predictive performance.

METHODS

This is a national cohort study including 22,448 patients who were older than 75 years and visited the hospital for emergency surgery from the cohort of older patients among the retrieved sample from the Korean National Health Insurance Service. The diagnostic and operation codes were one-hot encoded and entered into the predictive model using the extreme gradient boosting (XGBoost) as a machine learning technique. The predictive performance of the model for postoperative 90-day mortality was compared with those of previous frailty evaluation tools such as Operation Frailty Risk Score (OFRS) and Hospital Frailty Risk Score (HFRS) using the receiver operating characteristic curve analysis.

RESULTS

The predictive performance of the XGBoost, OFRS, and HFRS for postoperative 90-day mortality was 0.840, 0.607, and 0.588 on a c-statistics basis, respectively.

CONCLUSIONS

Using machine learning techniques, XGBoost to predict postoperative 90-day mortality, using diagnostic and operation codes, the prediction performance was improved significantly over the previous risk assessment models such as OFRS and HFRS.

Association of frailty with healthcare resource utilization after open thoracic/thoracolumbar posterior spinal fusion for adult spinal deformity

PURPOSE

The Hospital Frailty Risk Score (HFRS) is a frailty-identifying metric developed using ICD-10-CM codes. While other studies have examined frailty in adult spinal deformity (ASD), the HFRS has not been assessed in this population. The aim of this study was to utilize the HFRS to investigate the impact of frailty on outcomes in ASD patients undergoing posterior spinal fusion (PSF).

METHODS

A retrospective study was performed using the 2016-2019 National Inpatient Sample database. Adults with ASD undergoing elective PSF were identified using ICD-10-CM codes. Patients were categorized into HFRS-based frailty cohorts: Low (HFRS < 5) and Intermediate-High (HFRS ≥ 5). Patient demographics, comorbidities, intraoperative variables, and outcomes were assessed. Multivariate regression analyses were used to determine whether HFRS independently predicted extended length of stay (LOS), non-routine discharge, and increased cost.

RESULTS

Of the 7500 patients identified, 4000 (53.3%) were in the Low HFRS cohort and 3500 (46.7%) were in the Intermediate-High HFRS cohort. On average, age increased progressively with increasing HFRS scores (p < 0.001). The frail cohort experienced more postoperative adverse events (p < 0.001), greater LOS (p < 0.001), accrued greater admission costs (p < 0.001), and had a higher rate of non-routine discharge (p < 0.001). On multivariate analysis, Intermediate-High HFRS was independently associated with extended LOS (OR: 2.58, p < 0.001) and non-routine discharge (OR: 1.63, p < 0.001), though not increased admission cost (OR: 1.01, p = 0.929).

CONCLUSION

Our study identified HFRS to be significantly associated with prolonged hospitalizations and non-routine discharge. Other factors that were found to be associated with increased healthcare resource utilization include age, Hispanic race, West hospital region, large hospital size, and increasing number of AEs.

Comparison of the predictive ability of clinical frailty scale and hospital frailty risk score to determine long-term survival in critically ill patients: a multicentre retrospective cohort study

BACKGROUND

The Clinical Frailty Scale (CFS) is the most commonly used frailty measure in intensive care unit (ICU) patients. The hospital frailty risk score (HFRS) was recently proposed for the quantification of frailty. We aimed to compare the HFRS with the CFS in critically ill patients in predicting long-term survival up to one year following ICU admission.

METHODS

In this retrospective multicentre cohort study from 16 public ICUs in the state of Victoria, Australia between 1st January 2017 and 30th June 2018, ICU admission episodes listed in the Australian and New Zealand Intensive Care Society Adult Patient Database registry with a documented CFS, which had been linked with the Victorian Admitted Episode Dataset and the Victorian Death Index were examined. The HFRS was calculated for each patient using the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes that represented pre-existing conditions at the time of index hospital admission. Descriptive methods, Cox proportional hazards and area under the receiver operating characteristic (AUROC) were used to investigate the association between each frailty score and long-term survival up to 1 year, after adjusting for confounders including sex and baseline severity of illness on admission to ICU (Australia New Zealand risk-of-death, ANZROD).

RESULTS

7001 ICU patients with both frailty measures were analysed. The overall median (IQR) age was 63.7 (49.1-74.0) years; 59.5% (n = 4166) were male; the median (IQR) APACHE II score 14 (10-20). Almost half (46.7%, n = 3266) were mechanically ventilated. The hospital mortality was 9.5% (n = 642) and 1-year mortality was 14.4% (n = 1005). HFRS correlated weakly with CFS (Spearman's rho 0.13 (95% CI 0.10-0.15) and had a poor agreement (kappa = 0.12, 95% CI 0.10-0.15). Both frailty measures predicted 1-year survival after adjusting for confounders, CFS (HR 1.26, 95% CI 1.21-1.31) and HFRS (HR 1.08, 95% CI 1.02-1.15). The CFS had better discrimination of 1-year mortality than HFRS (AUROC 0.66 vs 0.63 p < 0.0001).

CONCLUSION

Both HFRS and CFS independently predicted up to 1-year survival following an ICU admission with moderate discrimination. The CFS was a better predictor of 1-year survival than the HFRS.

Hospital Frailty Risk Score Predicts Outcomes in Chronic Obstructive Pulmonary Disease Exacerbations

INTRODUCTION

Patients with chronic obstructive pulmonary disease (COPD) are at high risk for frailty and prone to complications after admission for an acute exacerbation. We aim to investigate the association between frailty risk and functional outcomes in patients with acute exacerbations of COPD, using a nationwide database.

METHODS

This retrospective cohort study included patients with acute exacerbations of COPD who were admitted by ambulance. We assessed frailty using the Hospital Frailty Risk Score (HFRS) and compared the outcomes between low frailty risk (HFRS < 5) and frailty at risk (HFRS ≥ 5) groups. The primary outcome was prolonged hospitalization (≥30 days). The secondary outcomes were in-hospital mortality, readmission (≤90 days), poor activities of daily living (ADL) at discharge, and difficulty in returning home.

RESULTS

There were 3,396 eligible patients (mean age, 75.9 ± 11.2 years; 20.4% female). The rate of frailty at risk patients was 14.0%. Frailty at risk patients were significantly higher rates of prolonged hospitalization (32.9% vs. 17.5%), more in-hospital mortality (16.4% vs. 12.5%), more difficulty in returning home (34.6% vs. 22.9%), and poorer ADL at discharge (8.7% vs. 12.4%) than those of low frailty risk. Multivariate analysis with adjusted covariates showed that HFRS was independently associated with prolonged hospitalization (odds ratio, 2.0; 95% confidence interval, 1.4-2.9).

CONCLUSIONS

HFRS can be used to predict the outcome of patients with acute exacerbations of COPD. This finding supports the validity of using the HFRS in clinical practice with patients with acute exacerbations of COPD.

Hospital Frailty Risk Score predicts adverse events in older patients with hip fractures after surgery: Analysis of a nationwide inpatient database in Japan

BACKGROUND

Frailty may predict adverse events in patients with hip fractures. This study aimed to investigate the association between frailty and adverse events in patients with hip fractures after surgery using information from Japanese health insurance.

METHODS

This retrospective cohort study included patients with hip fractures aged ≥ 65 years using a nationwide database in Japan. We examined the relationship of the frailty risk, which was defined using the Hospital Frailty Risk Score (HFRS), with in-hospital mortality, complications such as delirium and pneumonia, and functional outcomes. We used descriptive analysis, logistic regression, and linear regression analysis to estimate the association between the HFRS and outcomes in patients with hip fracture.

RESULTS

We analysed data from 36,192 patients with hip fractures after surgery (mean age: 83.6 ± 6.7 years, female: 79.5%). The proportions of low, intermediate, and high risk of frailty were 68.4%, 28.1%, and 3.5%, respectively. The frailty risk was independently associated with in-hospital mortality (intermediate risk: odds ratio [OR] 1.385, P < 0.001; high risk: OR 1.572; P < 0.001) and the occurrence of complications. Furthermore, each frailty risk was negatively associated with the Barthel Index score at discharge (intermediate risk: coefficient -11.919, P < 0.001; high risk: coefficient -18.044; P < 0.001).

CONCLUSIONS

The HFRS could predict adverse events, including in-hospital mortality, in Japanese older patients with hip fractures. This finding supports the validity of using the HFRS in clinical practice for patients with hip fractures.