204. The ALFA score predicts 30-day mortality after elective anterior lumbar interbody fusion

Abstract

BACKGROUND CONTEXT Anterior lumbar interbody fusion (ALIF) is an effective procedure for spondylolisthesis and pseudarthrosis but is not without risks. Surgeons and patients would benefit from an easy to use shared decision-making tool that can accurately predict patients at risk for mortality after ALIF. PURPOSE We developed the ALFA score to predict mortality after ALIF. ALFA is an acronym for age 65, loss of weight, functional dependence, and American Society of Anesthesiologists (ASA) physical status classification of 3. STUDY DESIGN/SETTING Retrospective review. PATIENT SAMPLE A total of 8,851 one or two level elective ALIF operations were retrieved from the National Surgical Quality Improvement Program American College of Surgeons (NSQIP-ACS) database from 2012 to 2018. Patients who had trauma, cancer, infection, spinal deformity, posterior arthrodesis, lateral approach, or were <18 years old were excluded. OUTCOME MEASURES Mortality. METHODS Bivariate analyses and multivariable logistic regression were employed to determine predictors of 30-day mortality. The ALFA score assigns points based on age 65 (3 point), functional dependence (5 points), ASA classification of 3 (4 points) and loss of weight >10% within 6 months preoperatively (10 points). Optimal tiers for the scoring system were determined using stratum-specific likelihood ratio (SSLR) analysis. The ALFA score was compared to two commonly utilized scores: the ASA physical status classification alone and the 5-factor modified frailty index (MFI-5). Area under the curve (AUC) analysis was employed to assess discriminative ability. The Hosmer-Lemeshow test was used to assess goodness of fit. Youden's Index was utilized to identify the optimal single score cut-off for the ALFA score to determine odds of 30-day mortality. Finally, the score was internally validated using a validation subset of patients from the ACS-NSQIP and externally validated using a cohort of patients undergoing ALIF from the Nationwide Inpatient Sample in 2019. RESULTS Nineteen patients (0.21%) did not survive for 30 days after ALIF. The AUC of the logistic regression model of the ALFA score was 0.79 on the internal training set and 0.80 on the internal validation set. This was more accurate than both the ASA physical status classification alone (AUC 0.73) and the mF-5 (AUC 0.67) with statistical significance (P < 0.05). SSLR analysis produced 4 distinct categories based on risk of mortality: 0.05% for score of 0, for 0.26% score of 1-4, 2.59% for score of 5-6, and 33.33% for score of 7+. Compared to an ALFA score of zero, an ALFA score of 1-4, 5-6, and 7+ had 5x, 50x, and 1000x greater odds of mortality (P < 0.05 for all). The optimal cut-off for the ALFA score was determined to be 4 (LR+ 3.6, J=0.49). The ALFA score had an AUC of 0.71 on external validation with NIS. The Hosmer-Lemeshow p value for goodness of fit was 0.10 indicating goodness of fit. CONCLUSIONS The ALFA score can be used to accurately predict patients who may be at risk for mortality within 30 days of ALIF. Surgeons can consider this easy to use and calculate tool to improve clinical care and shared decision making with patients. FDA DEVICE/DRUG STATUS This abstract does not discuss or include any applicable devices or drugs. Anterior lumbar interbody fusion (ALIF) is an effective procedure for spondylolisthesis and pseudarthrosis but is not without risks. Surgeons and patients would benefit from an easy to use shared decision-making tool that can accurately predict patients at risk for mortality after ALIF. We developed the ALFA score to predict mortality after ALIF. ALFA is an acronym for age 65, loss of weight, functional dependence, and American Society of Anesthesiologists (ASA) physical status classification of 3. Retrospective review. A total of 8,851 one or two level elective ALIF operations were retrieved from the National Surgical Quality Improvement Program American College of Surgeons (NSQIP-ACS) database from 2012 to 2018. Patients who had trauma, cancer, infection, spinal deformity, posterior arthrodesis, lateral approach, or were <18 years old were excluded. Mortality. Bivariate analyses and multivariable logistic regression were employed to determine predictors of 30-day mortality. The ALFA score assigns points based on age 65 (3 point), functional dependence (5 points), ASA classification of 3 (4 points) and loss of weight >10% within 6 months preoperatively (10 points). Optimal tiers for the scoring system were determined using stratum-specific likelihood ratio (SSLR) analysis. The ALFA score was compared to two commonly utilized scores: the ASA physical status classification alone and the 5-factor modified frailty index (MFI-5). Area under the curve (AUC) analysis was employed to assess discriminative ability. The Hosmer-Lemeshow test was used to assess goodness of fit. Youden's Index was utilized to identify the optimal single score cut-off for the ALFA score to determine odds of 30-day mortality. Finally, the score was internally validated using a validation subset of patients from the ACS-NSQIP and externally validated using a cohort of patients undergoing ALIF from the Nationwide Inpatient Sample in 2019. Nineteen patients (0.21%) did not survive for 30 days after ALIF. The AUC of the logistic regression model of the ALFA score was 0.79 on the internal training set and 0.80 on the internal validation set. This was more accurate than both the ASA physical status classification alone (AUC 0.73) and the mF-5 (AUC 0.67) with statistical significance (P < 0.05). SSLR analysis produced 4 distinct categories based on risk of mortality: 0.05% for score of 0, for 0.26% score of 1-4, 2.59% for score of 5-6, and 33.33% for score of 7+. Compared to an ALFA score of zero, an ALFA score of 1-4, 5-6, and 7+ had 5x, 50x, and 1000x greater odds of mortality (P < 0.05 for all). The optimal cut-off for the ALFA score was determined to be 4 (LR+ 3.6, J=0.49). The ALFA score had an AUC of 0.71 on external validation with NIS. The Hosmer-Lemeshow p value for goodness of fit was 0.10 indicating goodness of fit. The ALFA score can be used to accurately predict patients who may be at risk for mortality within 30 days of ALIF. Surgeons can consider this easy to use and calculate tool to improve clinical care and shared decision making with patients.