Sage Journals: Discover world-class research

Abstract

Study Design

Retrospective cohort study.

Objectives

Frailty and nutritional status are predictors of adverse spine surgery outcomes. This study evaluated the predictive utility of a combined Risk Analysis Index (RAI) and Geriatric Nutritional Risk Index (GNRI) model, and introduced a compound score integrating RAI, GNRI, American Society of Anesthesiologists (ASA) classification, and Preoperative Acute Severe Condition (PACS). To develop the CGARP score, we performed multivariable logistic regression with 30-day mortality as the dependent variable and GNRI, RAI, ASA, and PACS as independent variables.

Methods

Using the National Surgical Quality Improvement Program (NSQIP) database (2015-2020), we assessed predictive performance for mortality, morbidity, and length of stay in spine surgery patients. Demographics, comorbidities, and surgical risk factors were analyzed across RAI quartiles. Model performance was measured using area under receiver operating characteristic curve (AUROC).

Results

Among 360 133 patients, increasing frailty and malnutrition were independently associated with worse postoperative outcomes. The RAI-GNRI model showed outcome-specific discrimination, C-statistics 0.619 (reoperation) to 0.882 (mortality). The CGARP compound model outperformed individual predictors across all outcomes, with AUROCs of 0.882 (mortality), 0.762 (non-home discharge), 0.686 (extended length of stay), 0.694 (any complication), and 0.641 (readmission). Internal bootstrapping confirmed model stability. Random Forest was the most predictive machine-learning algorithm (AUC = 0.9553). Threshold analysis using Youden’s J statistic identified 4 risk categories, correlating with stepwise increases in mortality, complications, and non-home discharge.

Discussion

Frailty and nutritional risk are independently predictive of adverse spine surgery outcomes. The CGARP model demonstrated superior predictive performance and provides clinically actionable risk stratification.

Keywords

frailty risk analysis index geriatric nutritional risk index spine surgery outcomes

Introduction

Every year, approximately 900 000 adults in the United States undergo spine surgery for various pathological entities, including infections, neoplasms, deformity, and degeneration, aiming to enhance quality of life.^1,2 With an aging population, spine surgeries are increasingly common among patients over 65 years of age.³ Thus, judicious patient selection in older patients is critical, considering comorbidities, functional and nutritional status, and the growing recognition of frailty.

Frailty, or decreased baseline physiologic reserve, is an independent risk factor for adverse postoperative outcomes following spine surgery.^4-6 The Risk Analysis Index (RAI) has emerged as a leading frailty index, demonstrating strong predictive capability for spine surgery outcomes. Similarly, nutritional status has been evaluated using the Geriatric Nutritional Risk Index (GNRI), which correlates serum albumin and body weight.^7,8 In recent lumbar spine fusion studies, infection and mortality rates were linked to malnourishment. While both the RAI and GNRI are useful for predicting outcomes, their combined impact on risk stratification remains unexplored.

The present study sought to assess the impact of patient frailty and nutritional status using a combined RAI-GNRI model to improve risk stratification of patients undergoing spine surgery using a national, prospective registry.

Methods

Data Source and Patient Consent to Procedure

Patient data was extracted from the American College of Surgeons (ACS) National Surgical Quality Improvement Program (NSQIP) database from 2015 to 2020. This validated database, involving over 700 hospitals, collects more than 200 variables on surgical patients.⁹ The study was HIPAA-compliant and exempt from institutional review board approval. Informed consent was waived due to the retrospective design of the study.

Patient Selection

Current Procedural Terminology (CPT) codes were used to identify adult (18 years and older) patients in the ACS NSQIP database who underwent spine surgery between 2015 and 2020. Patients were identified using CPT codes 22 010-22855 and 62 380-63707 for spinal instrumentation and spinal procedures, respectively. Patients 90 years and older, as well as those who had missing outcome data (mortality, unknown discharge destination, etc.), unknown transfer status, and unknown functional status were excluded.^10-13

Risk Analysis Index (RAI)

The RAI was developed to optimize preoperative screening based on frailty stratification in patients undergoing elective surgery.¹⁴ The RAI evaluates a diverse set of patient factors and therefore more accurately captures the multidimensionality of frailty. The RAI incorporates variables such as age, sex, renal impairment, dyspnea, cancer status, weight loss, and functional status. Of note, cognitive deterioration, originally coded in the RAI scoring with Activities of Daily Living (ADL), was excluded in our study because preoperative cognitive decline is not recorded in the ACS NSQIP during the study period. Similarly, a few of the ACS NSQIP variables have been recently removed (eg, preoperative radiotherapy and chemotherapy). However, the absence of these variables does not significantly impact the model’s performance and therefore are conventionally ignored.^15,16 Patients in the present study were stratified into increasing frailty tiers: robust (RAI scores ≤20), normal (RAI scores 21-30), frail (RAI scores 31-40), and very frail (RAI scores ≥41).⁴

American Society of Anesthesiologists (ASA) Physical Status Classification

ASA is a risk assessment tool to preoperatively stratify patients based on overall health status. It contains 5 main categories: ASA I (healthy patient), ASA II (mild systemic disease), ASA III (severe systemic disease), ASA IV (severe systemic disease that is a constant threat to life), and ASA V (moribund patient with a guarded prognosis absent surgery). High ASA grades have been established in the literature as strongly correlating with poor postoperative outcomes across various surgical specialties, including spine surgery. ASA was thus additionally included as a component of the CGARP model.

Geriatric Nutritional Risk Index (GNRI)

The GNRI was originally developed using variables collected by the ACS-NSQIP database, including sex, weight (kg), height (cm), and preoperative albumin level. GNRI is calculated using the formula outlined by Bouillanne et al: GNRI = (1.489 × serum albumin [gram per liter]) + (41.7 × [weight/WLo]).¹⁷ The weight/WLo ratio was capped at 1 for patients with weight exceeding their ideal weight, in order to capture overweight/obese patients who may be malnourished. WLo is the ideal weight, calculated by the Devine formula: men as 50 + (0.91 × [height in centimeters −152.4]) and women as 45.5 + (0.91 × [height in centimeters −152.4]).

GNRI values were stratified into clinically relevant categories: no nutritional risk (GNRI scores >98), low risk (GNRI scores 92-98), moderate risk (GNRI scores 82-91), and major nutritional risk (GNRI scores <82).

Preoperative Acute Severe Condition (PACS)

PACS reflects the cumulative burden of acute preoperative conditions and is calculated by summing binary NSQIP “present at time of surgery” flags for sepsis, SIRS, acute renal failure, ventilator dependence, or pneumonia.

Combined GNRI-ASA-RAI-PACS (CGARP) Score

To develop the CGARP score, we performed multivariable logistic regression with 30-day mortality as the dependent variable and GNRI, RAI, ASA, and PACS as independent variables. The resulting adjusted odds ratios were used to derive weightings proportional to each variable’s relative contribution to risk. These coefficients were normalized and scaled for practical use, yielding the final formula: CGARP = (3 × GNRI group) + (2 × RAI group) + (2 × ASA class) + (1 × PACS). This approach allowed for data-driven, interpretable weighting while preserving clinical relevance.

Population Characteristics and Complications

The baseline study variables included in the present study are age, sex, BMI, hospital length of stay (LOS), and total operation time. Previous medical comorbidities including dyspnea, hypertension functional status, COPD, CHF, smoking status, DM, disseminated cancer, steroid use, weight loss, and bleeding disorders were also included. Preoperative Acute Severe Condition (PACS) was included as an additional risk variable, reflecting the cumulative burden of acute preoperative conditions. PACS is a simple variable that is the sum of all conditions a patient has on surgery, calculated using Present at Time of Surgery variables from the ACS NSQIP database.¹⁸

Outcomes and Complications

Outcomes of interest included 30-day mortality, overall complication rates, Clavien-Dindo grade IV (CDIV) complications, 30-day unplanned readmission, 30-day unplanned reoperation, extended LOS (eLOS), and nonhome discharge destination (NHD). Complications consisted of 1 of the following: prolonged intubation exceeding 48 hours, unplanned reintubation, sepsis, septic shock, pneumonia, deep vein thrombosis/thrombophlebitis, pulmonary embolism, acute cerebrovascular accident or stroke with neurological deficit, acute renal failure, myocardial infarction, cardiac arrest requiring cardiopulmonary resuscitation, superficial surgical site infection (SSI), deep incisional SSI, organ space SSI, wound disruption, intra- or postoperative blood transfusion, renal insufficiency, or urinary tract infection. CDIV complications are defined by the presence of life-threatening complications defined by single or multiple organ dysfunction requiring intensive care unit management.¹⁹ eLOS was defined as > 75th percentile of length of stay of all patients, which was equal to 4 days of stay or longer in our cohort.

Statistical Analysis

Non-normally distributed interval data were presented as median with Interquartile Range (IQR). The distribution of the data was assessed using the Kolmogorov-Smirnov test. Kruskal-Wallis test was utilized to analyze non-normally distributed continuous variables, while Chi-square was used to analyze categorical variables. Multivariable logistic regression analyses on ASA, GNRI, and RAI and 30-day mortality were carried out to arrive at the coefficients for the development of compound score index. Effect sizes for dichotomous outcomes were expressed as Odds Ratio (OR) with corresponding 95% Confidence Interval (CI). To ensure model stability and eliminate multicollinearity concerns, Variance Inflation Factor (VIF) was calculated for each input variable, with VIF values below commonly accepted thresholds (VIF <3) indicating low collinearity. Receiver Operating Characteristic (ROC) curve analysis was performed, incorporating Area Under the Curve (AUC)/C-statistics calculations to assess the discriminative ability ASA, GNRI, RAI, and the new compound score. The DeLong test was applied to analyze the differences in discrimination performances (C-statistic). Youden’s J statistic was employed to determine optimal thresholds for stratifying the CGARP into 4 distinct categories. Internal validation of the new compound score was carried out using 100 bootstrapping replications with replacement. The 4 CGARP component variables (RAI, GNRI, ASA, and PACS) were used as inputs to train multiple machine learning classifiers to assess their predictive value for 30-day postoperative adverse outcomes. Models included tree-based classifiers (Random Forest, XGBoost, and LightGBM), a kernel-based classifier (Support Vector Classifier), a neural network model, and ensemble methods (voting and stacking classifiers), in addition to a logistic regression model for baseline comparison. Predictive performance was assessed using 5-fold cross-validation and confusion matrix-derived metrics, with accuracy, precision, recall, and F1-score serving as primary evaluation criteria. Feature importance scores were extracted from tree-based models and normalized to enable direct comparison of the relative contribution of each CGARP component to outcome prediction. Statistical analyses were carried out using Excel (Microsoft, Redmond, WA, USA), IBM SPSS Statistics ver. 27.0 (IBM Co, Armonk, NY, USA), and STATA 12.0 (StataCorp, College Station, TX, USA). A P-value <0.05 was considered statistically significant.

Results

Patient Characteristics

A total of 382 789 patients underwent spine surgery during the study period, with 360 133 patients included in the analysis (Supplemental Figure 1). The cohort had a mean age of 60 years, was predominantly male (52.8%, N = 190 000) and white (76.4%, N = 275 184) (Table 1). Non-frail patients were significantly female (58.1%, P < 0.001). The percentage of non-white patients increased with an increase in frailty (P < 0.001). The most common diagnoses and procedures in our cohort were summarized (Supplemental Table 1, Supplemental Table 2).

Table 1.

The Association of Patient Demographics and Comorbidities and Risk Analysis Index (RAI) Tiers. COPD, Chronic Obstructive Pulmonary Disease; CHF, Congestive Heart Failure; mFI-5, Modified Frailty Index-5; RAI, Risk Analysis Index; GNRI, Geriatric Nutritional Risk Index

Variable	Total (n = 360 133)	Not frail (RAI ≤20) (n = 215 209)	Prefrail (RAI = 21-30) (n = 133 164)	Frail (RAI = 31-40) (n = 10 592)	Severely frail (RAI ≥41) (n = 1168)	P-value
Age (yr)	60 (40-96)	52(43-58)	71 (65-76)	74 (65-84)	70 (61-78)	<0.001
Sex, male	190 000 (52.8%)	90 228 (41.9%)	91 714 (68.9%)	7208 (68.1%)	850 (72.8%)	<0.001
White	275 184 (76.4%)	161 069 (74.8%)	106 061 (79.6%)	7320 (69.1%)	734 (62.8%)	<0.001
Non-white/Unknown	84 949 (23.6%)	54 140 (25.1%)	2703 (20.4%)	3272 (30.9%)	434 (37.1%)
Body mass index (kg/m²)	29.76 (26.00-34.20)	30.13 (26.1434.95)	26.97 (22.31-33.37)	25.71 (24.25-27.52)	25.00 (22.31-30.01)	<0.001
Functional status						<0.001
Partially dependent	8136 (2.3%)	827 (0.4%)	3840 (2.9%)	2938 (27.7%)	531 (45.5%)
Totally dependent	1010 (0.3%)	74 (0.0%)	198 (0.1%)	461 (4.4%)	277 (23.7%)
Diabetes mellitus						<0.001
Oral medication	22 564 (6.3%)	10 124 (4.7%)	11 114 (8.3%)	1173 (11.1%)	153 (13.1%)
Insulin	43 950 (12.2%)	19 864 (9.2%)	22 454 (16.9%)	1510 (14.3%)	122 (10.4%)
COPD	15 572 (4.3%)	6728 (3.1%)	7860 (5.9%)	857 (8.1%)	127 (10.9%)	<0.001
CHF	1470 (0.4%)	69 (0.0%)	823 (0.6%)	479 (4.5%)	99 (8.5%)	<0.001
Current smoker	73 610 (20.4%)	55 589 (25.8%)	16 508 (12.4%)	1277 (12.1%)	236 (20.2%)	0.002
Dyspnea at rest	1006 (0.3%)	196 (0.1%)	565 (0.4%)	186 (1.8%)	59 (5.1%)	0.114
Hypertension	185 881 (51.6%)	83 851 (39.0%)	94 324 (70.8%)	6988 (66.0%)	718 (61.5%)	<0.001
Disseminated cancer	5161 (1.4%)	0 (0.0%)	103 (0.1%)	4221 (39.9%)	837 (71.7%)	0.005
Steroid use	15 725 (4.4%)	7839 (3.6%)	6486 (4.9%)	1207 (11.4%)	193 (16.5%)	0.005
Weight loss	1671 (0.5%	121 (0.1%)	573 (0.4%)	528 (5.0%)	449 (38.4%)	<0.001
Residence other than independent living	1444 (0.4%)	232 (0.1%)	678 (0.5%)	420 (4.0%)	114 (9.8%)	<0.001
mFI-5						<0.001
Not frail (mFI-5 = 0)	155 396 (43.1%)	121 075 (56.3%)	31 755 (23.8%)	2418 (22.8%)	148 (12.7%)
Prefrail (mFI-5 = 1)	137 966 (38.3%)	68 162 (31.7%)	65 755 (49.4%)	3703 (35.0%)	346 (29.6%)
Frail (mFI-5 = 2)	60 191 (16.7%)	24 547 (11.4%)	32 313 (24.3%)	2930 (27.7%)	401 (34.3%)
Severely frail (mFI-5 ≥ 3)	6580 (1.8%)	1425 (0.7%)	3341 (2.6%)	1541 (14.6%)	273 (23.4%)
GNRI						<0.001
>98	123 046 (77.9%)	73 094 (82.6%)	46 749 (75.8%)	3011 (44.0%)	192 (20.2%)
92-98	20 799 (13.2%)	10 161 (11.5%)	8943 (14.5%)	1485 (21.7%)	210 (22.1%)
82-91	9480 (6.0%)	3482 (3.9%)	4177 (6.8%)	1478 (21.6%)	343 (36.1%)
<82	4637 (2.9%)	1736 (2.0%)	1834 (3.0%)	862 (12.6%)	205 (21.6%)
ASA						<0.001
I	13 969 (3.9%)	12 894 (6.0%)	1054 (0.8%)	20 (0.2%)	1 (0.1%)
II	165 334 (45.9%)	117 405 (54.6%)	46 429 (34.9%)	1400 (13.2%)	100 (8.6%)
III	168 702 (46.9%)	81 342 (37.8%)	79 462 (59.7%)	7237 (68.4%)	661 (56.6%)
IV	11 726 (3.3%)	3334 (1.6%)	6080 (4.6%)	1907 (18.0%)	405 (34.7%)
Outpatient procedure	111 739 (31.0%)	74 897 (34.8%)	35 722 (26.8%)	1093 (10.3%)	27 (2.3%)	<0.001
Length of stay after operation (day)	2 (1-3)	1 (1-3)	2 (1-4)	4 (2-8)	7 (4-11)	<0.001
Operative time (min)	125 (82-192)	119 (78-184)	134 (88-202)	156 (103-232)	167 (114-255)	0.100

RAI and GNRI 30-Day Outcomes

Univariate analysis revealed that severely frail patients had the highest incidence of mortality (12%, P < 0.001), non-routine discharge (66%, P < 0.001), eLOS (83.4%, P < 0.001), complications (49%, P < 0.001), Clavien-Dindo grade IV complications (15.6%, P < 0.001), readmission (19%, P < 0.001) and reoperation (9.3%, P < 0.001) (Table 2). Similar trends were observed in the association between GNRI and the outcomes. The highest rates of mortality (6.1%, P < 0.001), non-routine discharge (58.3%, P < 0.001), eLOS (73.4%, P < 0.001), complication (44.3%, P < 0.001), Clavien-Dindo grade IV complications (19.5%, P < 0.001), readmission (12.1%, P < 0.001), and reoperation (10.9%, P < 0.001) were observed in the group with major nutrition index (GNRI<82) (Table 3; Figure 1).

Table 2.

30-Day Outcome Measures Including Mortality, Nonroutine Discharge, Extended Length of Stay (eLOS), Occurrence of Complication, Clavien-Dindo Grade IV Complications, Reoperation, and Readmission Tiers

Variable	Total	Not frail (RAI ≤20)	Prefrail (RAI = 21-30)	Frail (RAI = 31-40)	Severely frail (RAI ≥41)	P value
Mortality	1440 (0.4%)	232 (0.1%)	619 (0.5%)	449 (4.2%)	140 (12.0%)	<0.001
Nonroutine discharge destination (Rehab, SNF, and Hospice, AMA)	49 512 (13.8%)	16 391 (7.6%)	27 368 (20.6%)	5002 (47.9%)	751 (65.8%)	<0.001
eLOS	84 587 (23.5%)	40 000 (18.6%)	37 411 (28.2%)	6232 (59.7%)	944 (83.4%)	<0.001
Any complication	34 741 (9.6%)	15 548 (7.2%)	15 473 (11.6%)	3150 (29.7%)	570 (48.8%)	<0.001
Clavien-Dindo grade IV	7725 (2.1%)	3113 (1.4%)	3592 (2.7%)	838 (7.9%)	182 (15.6%)	<0.001
Readmission	17 706 (4.9%)	8369 (3.9%)	7768 (5.8%)	1352 (12.8%)	217 (18.6%)	<0.001
Reoperation	10 651 (3.0%)	5691 (2.6%)	4239 (3.2%)	612 (5.8%)	109 (9.3%)	<0.001

Table 3.

30-Day Outcome Measures Including Mortality, Nonroutine Discharge, Extended Length of Stay (eLOS), Occurrence of Complication, Clavien-Dindo Grade IV Complications, Reoperation, and Readmission Among Geriatric Nutritional Risk Index (GNRI) Tiers

Variable	Total	GNRI >98	GNRI = 92-98	GNRI = 82-91	GNRI <82	P Value
Mortality	1068 (0.7%)	300 (0.2%)	186 (0.9%)	293 (3.0%)	289 (6.1%)	<0.001
Nonroutine discharge destination (Rehab, SNF, and Hospice, AMA)	28 993 (18.3%)	16 168 (13.1%)	5577 (26.6%)	4592 (48.2%)	2556 (58.3%)	<0.001
eLOS	44 837 (28.3%)	28 354 (22.9%)	7498 (35.8%)	5662 (59.6%)	3323 (73.4%)	<0.001
Any complication	19 418 (12.2%)	11 189 (9.0%)	3232 (15.4%)	2895 (30.0%)	2102 (44.3%)	<0.001
Clavien-Dindo grade IV	4810 (3.0%)	2064 (1.7%)	823 (3.9%)	997 (10.3%)	926 (19.5%)	<0.001
Readmission	9501 (6.0%)	6113 (4.9%)	1619 (7.7%)	1194 (12.4%)	575 (12.1%)	<0.001
Reoperation	5466 (3.4%)	3441 (2.8%)	863 (4.1%)	645 (6.7%)	517 (10.9%)	<0.001

Figure 1.

Rates of Adverse Outcomes Stratified by Categories of Geriatric Nutritional Risk Index (GNRI; Top) and Risk Analysis Index (RAI; Bottom). CDVI; Clavien-Dindo Class IV, eLOS; Extended Length of Stay

Major Postoperative Outcomes

A univariate logistic regression analysis examined the impact of GNRI and RAI on postoperative outcomes in spine surgery patients (Table 4). The findings indicate that lower GNRI and higher RAI scores are associated with significantly increased odds of adverse outcomes. Specifically, patients with the lowest GNRI (<82) and the highest RAI (≥41) have the highest odds ratios across all outcomes, with mortality odds ratios of 26.737 and 126.194, respectively. This trend is consistent across other measured outcomes, including nonroutine discharge, extended length of stay (eLOS), complications, Clavien-Dindo grade IV complications, readmissions, and reoperations.

Table 4.

Univariate Logistic Regression Analysis of GNRI and RAI and Major Postoperative Measures in Spine Surgery Patients. GNRI; Geriatric Nutritional Risk Index, RAI; Risk Analysis Index. Patient Groups With GNRI >98 and RAI ≤20 Were the Reference for GNRI and RAI Regression Analyses, Respectively

Outcome	GNRI category	Odds ratio (95%confidence interval)	RAI category	Odds ratio (95%confidence interval)
Mortality	92-98	3.672 (3.056-4.412)	21-30	4.327 (3.721-5.033)
	82-91	12.893 (10.961-15.166)	31-40	41.019 (34.964-48.123)
	<82	26.737 (22.687-31.512)	≥41	126.194 (101.423-157.015)
Nonroutine discharge destination	92-98	2.414 (2.331-2.500)	21-30	3.142 (3.078-3.208)
	82-91	6.194 (5.931-6.470)	31-40	11.117 (10.664-11.589)
	<82	9.280 (8.730-9.865)	≥41	23.252 (20.555-26.302)
eLOS	92-98	1.876 (1.818-1.935)	21-30	1.715 (1.687-1.743)
	82-91	4.963 (4.754-5.182)	31-40	6.484 (6.226-6.752)
	<82	9.288 (8.684-9.934)	≥41	21.952 (18.764-25.681)
Any complication	92-98	1.827 (1.751-1.905)	21-30	1.688 (1.649-1.728)
	82-91	4.314 (4.113-4.525)	31-40	5.436 (5.198-5.684)
	<82	8.022 (7.551-8.523)	≥41	12.240 (10.901-13.744)
Clavien-Dindo grade IV	92-98	2.401 (2.212-2.607)	21-30	1.889 (1.799-1.983)
	82-91	6.797 (6.283-7.353)	31-40	5.853 (5.409-6.334)
	<82	14.324 (13.171-15.579)	≥41	12.576 (10.695-14.788)
Readmission	92-98	1.605 (1.516-1.699)	21-30	1.531 (1.483-1.580)
	82-91	2.719 (2.546-2.904)	31-40	3.618 (3.403-3.846)
	<82	2.660 (2.429-2.913)	≥41	5.650 (4.867-6.558)
Reoperation	92-98	1.496 (1.387-1.614)	21-30	1.210 (1.163-1.260)
	82-91	2.506 (2.298-2.733)	31-40	2.258 (2.072-2.460)
	<82	4.284 (3.886-4.722)	≥41	3.789 (3.106-4.623)

Novel Score

A novel score (Combined GNRI-ASA-RAI-PACS; CGARP) was developed including GNRI, ASA, RAI, and PACS based on the results of multivariable regression of mortality (Table 5). The weighting scheme for CGARP was determined based on the adjusted odds ratios from multivariate logistic regression predicting 30-day mortality. The final coefficients were scaled to balance the relative contribution of each factor to overall risk stratification. Novel Score (CGARP ) = (3* GNRI) + (2*RAI) + (2*ASA) + (1*PACS). VIF and tolerance values were calculated for all covariates included in the CGARP model to assess multicollinearity (Table 6). All VIF values were well below the conservative threshold of 3, ranging from 1.13 to 1.22, and tolerance values exceeded 0.8 for all predictors. These findings indicate low collinearity among RAI, GNRI, PACS, and ASA, supporting their concurrent inclusion in multivariable and machine learning models.

Table 5.

Multivariable Regression Analysis of Mortality and American Society of Anesthesiologists Physical Status Class Risk Stratification System (ASA), Geriatric Nutritional Risk Index (GNRI), Risk Analysis Index (RAI), and Preoperative Acute Severe Condition (PACS)

Variable	Adjusted odds ratio	95% confidence interval - lower bound	95% confidence interval - upper bound	P-value
ASA	1.941	1.87	2.015	<0.001
PASC	1.29	1.206	1.379	<0.001
GNRI	2.581	2.528	2.635	<0.001
RAI	1.558	1.512	1.606	<0.001

Table 6.

Collinearity Diagnostics for Multivariable Model Covariates in CGARP

Variable	VIF	Tolerance
RAI	1.2	0.837
GNRI	1.16	0.862
PACS	1.13	0.885
ASA	1.22	0.819

Main Results

The performance of RAI and other indices were acceptable for most outcomes; however, the CGARP demonstrated a significantly superior performance to all individual indices (Table 7, Figures 2 and 3). Notably, RAI (C-statistic:0.833) and GNRI (C-statistic:0.817) had excellent predictive performance for mortality, while CGARP had a C-statistic of 0.882 which was distinctly superior.

Table 7.

AUC With 95% Confidence Interval for the American Society of Anesthesiologists (ASA) Physical Status Class Risk Stratification System (ASA), Geriatric Nutritional Risk Index (GNRI), Risk Analysis Index (RAI), and Combined GNRI-ASA-RAI-PACS (CGARP) and Post-operative Outcomes. The DeLong Test was Used to Compare all Indices Against the Novel Compound Score. AUC; Area Under the Receiver Operating Characteristic Curve

Outcome variable	Index	AUC	95 % confidence interval		P-value
Outcome variable	Index	AUC	Lower	Upper	P-value
Mortality	CGARP	0.882	0.871	0.894	Ref
	RAI	0.833	0.819	0.847	<0.001
	ASA	0.776	0.762	0.790	<0.001
	GNRI	0.817	0.802	0.831	<0.001
	mFI-5	0.658	0.641	0.676	<0.001
Nonroutine discharge destination	CGARP	0.762	0.759	0.765	Ref
	RAI	0.716	0.713	0.720	<0.001
	ASA	0.677	0.674	0.680	<0.001
	GNRI	0.687	0.684	0.691	<0.001
	mFI-5	0.625	0.622	0.629	<0.001
eLOS	CGARP	0.686	0.683	0.689	Ref
	RAI	0.626	0.623	0.629	<0.001
	ASA	0.636	0.633	0.639	<0.001
	GNRI	0.640	0.637	0.644	<0.001
	mFI-5	0.574	0.570	0.577	<0.001
Any complication	CGARP	0.694	0.690	0.699	Ref
	RAI	0.630	0.625	0.634	<0.001
	ASA	0.641	0.637	0.646	<0.001
	GNRI	0.653	0.649	0.658	<0.001
	mFI-5	0.577	0.572	0.581	<0.001
Clavien-Dindo class IV complications	CGARP	0.713	0.707	0.718	Ref
	RAI	0.614	0.608	0.620	<0.001
	ASA	0.657	0.652	0.663	<0.001
	GNRI	0.685	0.678	0.691	<0.001
	mFI-5	0.602	0.594	0.611	<0.001
Readmission	CGARP	0.641	0.635	0.647	Ref
	RAI	0.596	0.590	0.603	<0.001
	ASA	0.605	0.599	0.610	<0.001
	GNRI	0.599	0.593	0.605	<0.001
	mFI-5	0.576	0.570	0.582	<0.001
Reoperation	CGARP	0.619	0.611	0.627	Ref
	RAI	0.554	0.546	0.562	<0.001
	ASA	0.593	0.586	0.601	<0.001
	GNRI	0.599	0.591	0.607	<0.001
	mFI-5	0.546	0.538	0.544	<0.001

Figure 2.

Receiver Operating Characteristic Curve of the American Society of Anesthesiologists Physical Status Class Risk Stratification System (ASA), Geriatric Nutritional Risk Index (GNRI), Risk Analysis Index (RAI), and the Compound Score for Major Postoperative Outcomes. Mortality (A), Discharge to Non-routine Destination (B), Extended Length of Stay (eLOS) (C), Any Complication (D), Clavien-Dindo Physical Status Classification IV Complication (E), Readmission (F), and Reoperation (G) in Patients Who had Spine Surgery From National Surgical Quality Improvement Program Database 2015-2020

Figure 3.

Forrest Plot of Area Under the Receiver Operating Characteristic Curve (AUC) of the American Society of Anesthesiologists Physical Status Class Risk Stratification System (ASA), Geriatric Nutritional Risk Index (GNRI), Risk Analysis Index (RAI), and Combined GNRI-ASA-RAI-PACS (CGARP) for Major Postoperative Outcomes in Patients Who had Spine Surgery. Extended Length of Stay; eLOS

Using Youden’s J statistic, CGARP scores were stratified into 4 clinically interpretable risk groups: Low Risk (≤15), Medium Risk (16-20), High Risk (21-25), and Very High Risk (>25). These thresholds reflect inflection points in the predictive utility of the model. Table 8 summarizes outcome rates across these categories. Patients in the Low Risk group experienced a 0.1% 30-day mortality rate, 8.4% rate of NHD, and a 7.2% complication rate. In contrast, patients in the Very High Risk group had a 14.0% mortality rate, 72.4% NHD, and a 55.8% complication rate. The Very High Risk group also exhibited the highest rates of CDIV complications (20.6%), unplanned readmission (16.4%), and unplanned reoperation (11.1%).

Table 8.

Postoperative Outcome Rates Stratified by CGARP Risk Category Based on Youden’s J Statistic

Risk category	Mortality %	Non-home discharge %	eLOS %	Complication %	CDIV complication %	Reoperation %	Readmission %
Low risk (≤15)	0.10	8.40	19.20	7.20	1.20	2.50	4.00
Medium risk (16-20)	0.50	26.60	35.20	15.00	3.80	4.00	7.70
High risk (21-25)	3.80	57.80	67.20	35.30	12.40	7.70	13.40
Very high risk (>25)	14.00	72.40	87.30	55.80	20.60	11.10	16.40

Machine Learning Model Performance Comparison

Performance metrics for individual and ensemble machine learning classifiers using the 4 CGARP variables as predictors are presented in Table 9. Among individual models, the Random Forest Classifier achieved the highest accuracy (0.9553) but had relatively low recall (0.3049). In contrast, the Neural Network model demonstrated balanced performance with high accuracy (0.833) and recall (0.8293). Among ensemble methods, the Voting Classifier (combining Random Forest, Support Vector Classifier, and XGBoost) yielded the highest F1-score (0.0911), indicating improved balance between precision and recall over individual models. Logistic regression, included for baseline comparison, performed modestly with an F1-score of 0.061.

Table 9.

Performance Comparison of Individual and Ensemble Machine Learning Models for CGARP Prediction

Model	Accuracy	Precision	Recall	F1 score
Logistic regression	0.8209	0.0317	0.8171	0.061
Neural network	0.833	0.0344	0.8293	0.0661
Tree-based classifiers
Random forest classifier	0.9553	0.0518	0.3049	0.0885
XGBoost	0.8766	0.0442	0.7927	0.0838
LightGBM (proxy gbm)	0.8546	0.0383	0.8049	0.0731
Kernel-based classifiers
Support vector classifier	0.8718	0.0427	0.7927	0.0809
Ensemble methods
Voting classifier (RF + SVM + XGB)	0.8926	0.0485	0.7561	0.0911
Stacking classifier (RF + SVM + XGB → GLM)	0.8884	0.0469	0.7644	0.0886

Variable Importance Analysis

Feature importance scores were extracted from tree-based classifiers and normalized for direct comparison (Figure 4). The Risk Analysis Index (RAI) demonstrated the highest predictive contribution, followed by GNRI, ASA, and PACS. These results underscore the dominant influence of RAI and GNRI in CGARP-based predictive modeling of postoperative outcomes.

Figure 4.

Variable Importance Scores for RAI, GNRI, ASA, and PACS in CGARP

Internal Validation

To provide bias-corrected performance of this CGARP novel index and evaluate its robustness, stability, and generalizability, bootstrap sampling for internal validation was performed (Table 10). The initial AUC values indicate the model’s performance before validation, with mortality having the highest initial AUC of 0.882, suggesting excellent predictive ability. After internal validation, the AUC for mortality slightly decreased to 0.864, remaining high, with a bias-corrected confidence interval between 0.8498 and 0.8790. Other outcomes, such as nonroutine discharge destination, extended length of stay (eLOS), any complication, readmission, and reoperation, showed lower initial and validated AUCs, indicating moderate to poor predictive performance. Notably, Clavien-Dindo grade IV complications maintained a stable AUC around 0.712, with a tight confidence interval, reflecting consistent predictive accuracy for severe complications.

Table 10.

Internal Validation of Area Under the Receiver Operating Curve Analysis of American Society of Anesthesiologists Physical Status Class Risk Stratification System (ASA), Geriatric Nutritional Risk Index (GNRI), Risk Analysis Index (RAI), and Combined GNRI-ASA-RAI-PACS (CGARP) by Bootstrapping Replications

Outcome variable	Index	Initial AUC	Internal validation AUC	Bias-corrected confidence intervals
Outcome variable	Index	Initial AUC	Internal validation AUC	Lower bound	Upper bound
Mortality	CGARP	0.882	0.864	0.8498	0.8790
	RAI	0.833	0.820	0.806	0.833
	ASA	0.776	0.614	0.597	0.626
	GNRI	0.817	0.804	0.789	0.817
	mFI-5	0.658	0.518	0.500	0.534
Nonroutine discharge destination (Rehab, SNF, and Hospice, AMA)	CGARP	0.762	0.685	0.681	0.689
	RAI	0.716	0.696	0.692	0.699
	ASA	0.677	0.529	0.525	0.532
	GNRI	0.687	0.664	0.661	0.669
	mFI-5	0.625	0.531	0.526	0.534
eLOS	CGARP	0.686	0.610	0.589	0.614
	RAI	0.626	0.604	0.602	0.607
	ASA	0.636	0.574	0.571	0.577
	GNRI	0.640	0.615	0.612	0.619
	mFI-5	0.574	0.549	0.546	0.551
Any complication	CGARP	0.694	0.624	0.620	0.628
	RAI	0.630	0.608	0.605	0.613
	ASA	0.641	0.566	0.562	0.569
	GNRI	0.653	0.630	0.625	0.635
	mFI-5	0.577	0.561	0.557	0.566
Clavien-Dindo grade IV complications	CGARP	0.713	0.712	0.704	0.719
	RAI	0.614	0.613	0.603	0.620
	ASA	0.657	0.510	0.504	0.523
	GNRI	0.685	0.709	0.700	0.717
	mFI-5	0.602	0.551	0.540	0.561
Readmission	CGARP	0.641	0.570	0.563	0.575
	RAI	0.596	0.575	0.569	0.580
	ASA	0.605	0.583	0.578	0.588
	GNRI	0.599	0.574	0.569	0.579
	mFI-5	0.576	0.554	0.546	0.561
Reoperation	CGARP	0.619	0.551	0.543	0.561
	RAI	0.554	0.532	0.523	0.538
	ASA	0.593	0.570	0.561	0.576
	GNRI	0.599	0.575	0.567	0.583
	mFI-5	0.546	0.535	0.526	0.545

Discussion

In this study, the RAI, GNRI, PACS and ASA indices were evaluated for both individual and combined use as a novel compound score, the CGARP (Combined GNRI-ASA-RAI-PACS), to predict postoperative adverse outcomes in spine surgery NSQIP patients from 2015-2020. This study contributes to a more robust understanding of the intersectionality between patient frailty and nutritional status, underscoring the importance of frailty and nutritional status in informing spine surgery candidate selection and guiding perioperative risk stratification.

Interpretation of Results

Our dataset consisted of older patients with a median age of 60 years. Notably, this advanced age is associated with increased frailty levels, which complicates perioperative recovery and heightens unfavorable outcomes.^5,6,20-22 Moreover, the compound score revealed that frail patients have the worst outcomes on all fronts—accounting for mortality, non-routine discharge, length of stay, complications, readmissions, and reoperations—compared with patients who are non-frail and those who are only at risk for being frail. These findings are in line with dozens of recent publications by our research group and by others demonstrating frailty as an independent predictor of adverse outcomes after brain and/or spine surgery^22-24 A significant gender discrepancy was also noted, with males comprising 53% of the cohort and being more frail on average than females. Males had a higher risk of worse postoperative outcomes—including longer hospital stays, higher complication rates, and increased mortality. Underlying reasons for the sex differences in frailty among the original cohort could include different body compositions, as well as differing health-related behaviors not directly measured by RAI or GNRI including smoking, alcohol use, or obesity status that have been strongly associated with frailty and adverse spine surgery outcomes.^25-30

Significant disparities across race and body weights were also observed in frailty. Patients who tended to be frail and required hospitalization for their surgical care had lower average BMI and tended to be non-white. These findings may emphasize the intersection of nutritional status and sociodemographic factors that are a critical driver of existing health disparities for both minority and frail patients.^31,32

Implications of the CGARP Score

Interestingly, the application of the compound CGARP score showed superior prognostic ability in predicting discharge destination than the use of any 1 individual metric. The consistency of the composite score in predicting outcomes such as mortality is consistent with a theoretical framework of frailty: by representing more approaches to measuring frailty across these available tools, there is a more granular risk characterization compared to analysis via any 1 tool alone. The weighting of CGARP was intentionally derived from regression-based risk estimates, and sensitivity analyses confirmed that equal weighting resulted in diminished discriminatory performance. This suggests that the differential impact of frailty, nutritional status, and surgical risk factors should be accounted for in composite risk stratification models.^33-35

The new composite score using RAI, GNRI, PACS, and ASA predicts postoperative outcomes with higher accuracy than any of the individual indices alone. GNRI alone predicted mortality, nonroutine discharge, eLOS, complications, Clavien-Dindo grade IV, readmission, and reoperation. RAI predicted mortality and Clavien-Dindo grade IV complications more accurately than ASA. The composite score showed stable prediction accuracy after internal validation, covering a wider scope of risk factors for better risk stratification and comprehensive patient management. The higher incidence of invasive surgery in this population necessitates the importance of accurate risk management. The greater gap between composite and individual scores highlights the improved discriminatory power of the composite score. Bowers et al demonstrated that the RAI-rev outperforms the mFI-5 in predicting 30-day mortality, Clavien-Dindo grade IV complications, nonhome discharge, and reoperation after ACDF. Covell et al found that the mASD-FI is effective for preoperative risk assessment in ASD surgery, predicting longer lengths of stay, higher complication rates, and inferior postoperative outcomes. While these studies have continued to validate the RAI as the premier frailty scale in spine surgery, improvements in outcome prognostication through frailty and nutritional status quantification may be a natural next step in risk stratification models.

The risk stratification table based on CGARP thresholds provides a practical and intuitive framework for clinical implementation. By categorizing patients into 4 tiers—Low, Medium, High, and Very High Risk—this analysis translates complex predictive modeling into accessible decision-making tools for clinicians. The identified cutoffs demonstrate a clear stepwise increase in the incidence of mortality, complications, and resource-intensive outcomes, such as NHD and extended LOS, across escalating CGARP strata.³⁶ These gradients not only validate the discriminatory accuracy of CGARP but also support its potential role in preoperative counseling, triage for enhanced recovery after surgery (ERAS) protocols, and identifying candidates for prehabilitation or nonoperative management.³⁷ While preliminary, this stratification framework offers actionable guidance and a foundation for future validation efforts aimed at threshold optimization and integration into perioperative workflows.

Other studies have found similar results utilizing the RAI framework., Bowers et al revealed in other spine and neurosurgical contexts like metastatic brain tumor resections that a higher RAI score is associated with increased risks of 30-day mortality, Clavien-Dindo grade IV complications and non-routine discharges across.^5,30,38-40 Frailty scores can potentially be mitigated by preoperative identification, as surgeons may identify frail patients and tailor perioperative management to minimize associated risks. Similarly, the GNRI, demonstrates the impact of nutritional status on perioperative outcomes. Patients with malnutrition are at risk of higher rates of complications, longer lengths of stay and mortality. Moreover, preoperative nutritional supplementation is associated with reduced postoperative spine surgery complications.^1,41-43 RAI-GNRI combines aspects of disease and nutritional status to enable an improved degree of risk stratification that allows for personalized interventions and higher quality care for each patient.

The results of this investigation also demonstrate that malnutrition or suboptimal nutritional status reduces the already impaired and reduced physiological reserve that characterizes frail patients and increases the physiological stress of surgery. This can further impede recovery, resulting in increased length of hospital stays and complication rates. This association between diminished GNRI score and increased complications parallels similar findings in previous literature, reinforcing the essential role nutrition has in achieving a successful postoperative recovery.^20,24-27

Limitations

The present study has several limitations. The retrospective design may have subjective biases, resulting in inaccurate data entries or inconsistencies in the study parameters. We acknowledge that our analysis did not include calibration metrics or external validation, which limits its immediate clinical implementation. As such, we emphasize the need for future prospective validation in multicenter and non-NSQIP datasets. This would enhance generalizability and allow formal calibration analysis, such as calibration plots or Hosmer-Lemeshow tests, to better guide clinical decision-making. External validation of the CGARP score’s translational usefulness across various surgical groups and healthcare settings is essential. Therefore, future studies need to assess its effectiveness in real-world clinical settings and non-spine surgery populations to validate its wider application and scalability. Additionally, the large national database may not capture specific institutional protocols for measuring frailty indices, which could affect individual results. No tool can fully encompass the nuances of frailty and nutritional status. Although many of these indexes focus on functional ability, cognition, and exposure to factors predicted to increase mortality and morbidity, they predominantly assess frailty based on physical performance and health deficits. Nutritional assessments are primarily based on serum albumin and weight loss, overlooking the importance of specific nutrient adequacy and overall diet quality, which can significantly influence frailty and postoperative recovery.⁴⁴

Conclusion

This large data study reveals the synergistic effect of both frailty and impaired nutritional status in predicting postoperative adverse outcomes in patients undergoing spine surgery. The proposed compound CGARP score, which combines RAI, GNRI, PACS, and ASA classifications, demonstrated superior discrimination in predicting adverse outcomes in spine surgery. This innovation refines the ability to identify spine surgery patients at high-risk for adverse outcomes, allowing for personalized interventions to improve outcomes for these patients.

Supplemental Material

Supplemental Material - Predictive Value of Geriatric Nutritional Risk Index and Risk Analysis Index for Post-operative Outcomes in Spine Surgery Patients: A Comprehensive Analysis

Supplemental Material for Predictive Value of Geriatric Nutritional Risk Index and Risk Analysis Index for Post-operative Outcomes in Spine Surgery Patients: A Comprehensive Analysis by Stefan T. Prvulovic, BS, Sina Zoghi, MD, Aryan Gajjar, Cameron J. Sabet, Michael M. Covell, BA, Bhavya Pahwa, Nithin Gupta, BS, Meic H. Schmidt, MD, MBA, Marc D. Moisi, MD, Johnny Delashaw MD, and Christian A. Bowers, MD in Global Spine Journal.

Footnotes

ORCID iDs

Stefan T. Prvulovic

Sina Zoghi

Nithin Gupta

Christian A. Bowers

Ethics Considerations

IRB approval was not required for this study as all patient information provided by the ACS-NSQIP is de-identified and publicly available.

Author Contributions

STP: conceptualization, methodology, data collection, formal analysis, writing – original draft, writing – review & editing; SZ: data collection, formal analysis, writing – original draft; AG: writing – original draft; CS: writing; - original draft; MMC: writing – review & editing; BP: data collection; NG: data collection; MHS: writing – review & editing, supervision; MDM: writing – review & editing, supervision; JD: writing – review & editing, supervision; CAB: conceptualization, methodology, formal analysis, writing – review & editing, supervision.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

Data used to produce this manuscript are available from the corresponding author upon reasonable request.*

Presentation Disclaimer

The following findings were presented at the Congress of Neurological Surgeons 2024 Annual Meeting in Houston, Texas. These findings were shared during our 6-minute oral presentation in the Sunrise Session on October 1^st, 2024.

Supplemental Material

Supplemental material for this article is available online.

Appendix

References

Mendiratta

Para

Berg

Vives

. Use of the geriatric nutritional risk index to assess risk for postoperative complications following posterior cervical decompression/fusion. Int J Spine Surg. 2023;17(6):866-874. doi:10.14444/8551

McIntosh

Hall

Melles

. The incidence of spinal surgery in Canada. Can J Surg. 1998;41(1):59-66.

Zou

Yang

. The aging spine: china facing up to the global challenge. Int J Spine Surg. 2022;16(5):861-862. doi:10.14444/8317. Epub 2022 Oct 11.

Covell

Rumalla

Kassicieh

, et al. Frailty measured by risk analysis index and adverse discharge outcomes after adult spine deformity surgery: analysis of 3104 patients from a prospective surgical registry (2011-2020). Spine J. 2023;23(5):739-745. doi:10.1016/j.spinee.2022.12.014

Bowers

Varela

Conlon

, et al. Comparison of the risk analysis index and the modified 5-factor frailty index in predicting 30-day morbidity and mortality after spine surgery. J Neurosurg Spine. 2023;39(1):136-145. doi:10.3171/2023.2.SPINE221019

Owodunni

Yocky

Courville

, et al. A comprehensive analysis of the triad of frailty, aging, and obesity in spine surgery: the risk analysis index predicted 30-day mortality with superior discrimination. Spine J. 2023;23(12):1778-1789. doi:10.1016/j.spinee.2023.08.008

Bouillanne

Morineau

Dupont

, et al. Geriatric nutritional risk index: a new index for evaluating at-risk elderly medical patients. Am J Clin Nutr. 2005;82(4):777-783. doi:10.1093/ajcn/82.4.777

Sun

. Geriatric nutritional risk index predicts adverse outcomes in human malignancy: a meta-analysis. Dis Markers. 2019;2019:4796598. doi:10.1155/2019/4796598

Raval

Pawlik

. Practical guide to surgical data sets: National surgical quality improvement program (NSQIP) and pediatric NSQIP. JAMA Surg. 2018;153(8):764-765. doi:10.1001/jamasurg.2018.0486

10.

Pierce

Naessig

Kummer

, et al. The five-item modified frailty index is predictive of 30-day postoperative complications in patients undergoing spine surgery. Spine. 2021;46(14):939-943. doi:10.1097/BRS.0000000000003936

11.

Wilson

JRF

Badhiwala

Moghaddamjou

Yee

Wilson

Fehlings

. Frailty is a better predictor than age of mortality and perioperative complications after surgery for degenerative cervical myelopathy: an analysis of 41,369 patients from the NSQIP database 2010-2018. J Clin Med. 2020;9(11):3491. doi:10.3390/jcm9113491

12.

Kazim

Dicpinigaitis

Bowers

, et al. Frailty status is a more robust predictor than age of spinal tumor surgery outcomes: a NSQIP analysis of 4,662 patients. Neurospine. 2022;19(1):53-62. doi:10.14245/ns.2142770.385

13.

Yagi

Michikawa

Hosogane

, et al. The 5-Item modified frailty index is predictive of severe adverse events in patients undergoing surgery for adult spinal deformity. Spine. 2019;44(18):E1083-E1091. doi:10.1097/BRS.0000000000003063

14.

Hall

Arya

Schmid

, et al. Development and initial validation of the risk analysis index for measuring frailty in surgical populations. JAMA Surg. 2017;152(2):175-182. doi:10.1001/jamasurg.2016.4202

15.

George

Hall

Youk

, et al. Association between patient frailty and postoperative mortality across multiple noncardiac surgical specialties. JAMA Surg. 2021;156(1):e205152. doi:10.1001/jamasurg.2020.5152

16.

George

Massarweh

Youk

, et al. Comparing veterans affairs and private sector perioperative outcomes after noncardiac surgery. JAMA Surg. 2022;157(3):231-239. doi:10.1001/jamasurg.2021.6488. Erratum in: JAMA Surg. 2022 Mar 1;157(3):282. doi: 10.1001/jamasurg.2022.0011.

17.

Shiloach

Frencher

Jr Steeger

, et al. Toward robust information: data quality and inter-rater reliability in the American college of surgeons national surgical quality improvement program. J Am Coll Surg. 2010;210(1):6-16. doi:10.1016/j.jamcollsurg.2009.09.031

18.

Jacobs

Schmidt

Hall

, et al. Differentiating urgent from elective cases matters in minority populations: developing an ordinal “Desirability of Outcome Ranking” to increase granularity and sensitivity of surgical outcomes assessment. J Am Coll Surg. 2023;237(3):545-555. doi:10.1097/XCS.0000000000000776

19.

Clavien

Barkun

de Oliveira

, et al. The Clavien-Dindo classification of surgical complications: five-year experience. Ann Surg. 2009;250(2):187-196. doi:10.1097/SLA.0b013e3181b13ca2

20.

Agarwal

Goldschmidt

Taylor

, et al. Impact of frailty on outcomes following spine surgery: a prospective cohort analysis of 668 patients. Neurosurgery. 2021;88(3):552-557. doi:10.1093/neuros/nyaa468

21.

Mohamed

Ramachandran

Rabai

, et al. Frailty assessment and prehabilitation before complex spine surgery in patients with degenerative spine disease: a narrative review. J Neurosurg Anesthesiol. 2023;35(1):19-30. doi:10.1097/ANA.0000000000000787

22.

Kweh

Lee

Tan

, et al. Spinal surgery in patients aged 80 years and older: risk stratification using the modified frailty index. Glob Spine J. 2021;11(4):525-532. doi:10.1177/2192568220914877. Epub 2020 Mar 30.

23.

Kweh

BTS

Lee

Tan

, et al. Risk stratification of elderly patients undergoing spinal surgery using the modified frailty index. Glob Spine J. 2023;13(2):457-465. doi:10.1177/2192568221999650

24.

Baek

Park

Kim

. Impact of frailty on the outcomes of patients undergoing degenerative spine surgery: a systematic review and meta-analysis. BMC Geriatr. 2023;23(1):771. doi:10.1186/s12877-023-04448-2

25.

Mielke

Schneider

Huscher

Ebert

Schaeffner

. Gender differences in frailty transition and its prediction in community-dwelling old adults. Sci Rep. 2022;12(1):7341. doi:10.1038/s41598-022-11358-7

26.

Crow

Lohman

Titus

, et al. Association of obesity and frailty in older adults: NHANES 1999-2004. J Nutr Health Aging. 2019;23(2):138-144. doi:10.1007/s12603-018-1138-x

27.

Kojima

Jivraj

Iliffe

Falcaro

Liljas

Walters

. Alcohol consumption and risk of incident frailty: the English longitudinal study of aging. J Am Med Dir Assoc. 2019;20(6):725-729. doi:10.1016/j.jamda.2018.10.011

28.

Liu

Yang

, et al. Genetic predisposition to smoking in relation to the risk of frailty in ageing. Sci Rep. 2023;13(1):2405. doi:10.1038/s41598-023-28780-0

29.

Kojima

Iliffe

Walters

. Smoking as a predictor of frailty: a systematic review. BMC Geriatr. 2015;15:131. doi:10.1186/s12877-015-0134-9

30.

Conlon

Thommen

Kazim

, et al. Risk analysis index and its recalibrated version predict postoperative outcomes better than 5-Factor modified frailty index in traumatic spinal injury. Neurospine. 2022;19(4):1039-1048. doi:10.14245/ns.2244326.163

31.

Anand

Khurrum

Chehab

, et al. Racial and ethnic disparities in frail geriatric trauma patients. World J Surg. 2021;45(5):1330-1339. doi:10.1007/s00268-020-05918-z

32.

De la Garza Ramos

Choi

Naidu

, et al. Racial disparities in perioperative morbidity following oncological spine surgery. Glob Spine J. 2023;13(5):1194-1199. doi:10.1177/21925682211022290

33.

Elsamadicy

Koo

Reeves

, et al. Utilization of machine learning to model important features of 30-day readmissions following surgery for metastatic spinal column tumors: the influence of frailty. Glob Spine J. 2024;14(4):1227-1237. doi:10.1177/21925682221138053

34.

Passias

Moattari

Pierce

, et al. Performance of the modified adult spinal deformity frailty index in preoperative risk assessment. Spine. 2022;47(20):1463-1469. doi:10.1097/BRS.0000000000004342

35.

Bowers

Varela

Naftchi

, et al. Superior discrimination of the risk analysis index compared with the 5-item modified frailty index in 30-day outcome prediction after anterior cervical discectomy and fusion. J Neurosurg Spine. 2023;39(4):509-519. doi:10.3171/2023.5.SPINE221020

36.

Hadaya

Sanaiha

Juillard

Benharash

. Impact of frailty on clinical outcomes and resource use following emergency general surgery in the United States. PLoS One. 2021;16(7):e0255122. doi:10.1371/journal.pone.0255122

37.

Naftalovich

Singal

Iskander

. Enhanced Recovery After Surgery (ERAS) protocols for spine surgery – review of literature. Anaesthesiol Intensive Ther. 2022;54(1):71-79. doi:10.5114/ait.2022.113961

38.

Dicpinigaitis

Khamzina

Hall

, et al. Adaptation of the risk analysis index for frailty assessment using diagnostic codes. JAMA Netw Open. 2024;7(5):e2413166. doi:10.1001/jamanetworkopen.2024.13166

39.

Covell

Warrier

Rumalla

Dehney

Bowers

. RAI-Measured frailty predicts non-home discharge following metastatic brain tumor resection: national inpatient sample analysis of 20,185 patients. J Neuro Oncol. 2023;164(3):663-670. doi:10.1007/s11060-023-04461-w

40.

Yocky

Owodunni

Courville

, et al. The risk analysis index has superior discrimination compared with the modified frailty index-5 in predicting worse postoperative outcomes for the octogenarian neurosurgical patient. Neurosurg Pract. 2023;4(3):e00044. doi:10.1227/neuprac.0000000000000044

41.

Chen

Zhu

, et al. Preoperative geriatric nutritional risk index is useful factor for predicting postoperative delirium among elderly patients with degenerative lumbar diseases. Eur Spine J. 2024;33(3):1055-1060. doi:10.1007/s00586-024-08142-8. Epub 2024 Feb 11. PMID: 38341814.

42.

Huang

Wang

Liu

Zheng

. Do nutritional assessment tools (PNI, CONUT, GNRI) predict adverse events after spinal surgeries? A systematic review and meta-analysis. J Orthop Surg Res. 2024;19(1):289. doi:10.1186/s13018-024-04771-3

43.

Cho

Choi

Lee

. Preoperative nutritional status and postoperative health outcomes in older adults undergoing spine surgery: electronic health records analysis. Geriatr Nurs. 2024;57:103-108. doi:10.1016/j.gerinurse.2024.04.005

44.

Khalooeifard

Rahmani

Tavanaei

, et al. The effect of vitamin D deficiency on outcomes of patients undergoing elective spinal fusion surgery: a systematic review and meta-analysis. Int J Spine Surg. 2022;16(1):53-60. doi:10.14444/8177

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.10 MB