Sage Journals: Discover world-class research

Abstract

Objective

The Hospital Frailty Risk Score (HFRS) is a metric that measures frailty among patients in large national datasets using ICD-10 codes. While other metrics have been utilized to demonstrate the association between frailty and poor outcomes in spine oncology, none have examined the HFRS. The aim of this study was to investigate the impact of frailty using the HFRS on complications, length of stay, cost of admission, and discharge disposition in patients undergoing surgery for primary tumors of the spinal cord and meninges.

Methods

A retrospective cohort study was performed using the Nationwide Inpatient Sample database from 2016 to 2018. Adult patients undergoing surgery for primary tumors of the spinal cord and meninges were identified using ICD-10-CM codes. Patients were categorized into 2 cohorts based on HFRS score: Non-Frail (HFRS<5) and Frail (HFRS≥5). Patient characteristics, treatment, perioperative complications, LOS, discharge disposition, and cost of admission were assessed.

Results

Of the 5955 patients identified, 1260 (21.2%) were Frail. On average, the Frail cohort was nearly 8 years older ( P < .001 ) and experienced more postoperative complications ( P = .001 ). The Frail cohort experienced longer LOS ( P < .001 ), a higher rate of non-routine discharge ( P = .001 ), and a greater mean cost of admission ( P < .001 ). Frailty was found to be an independent predictor of extended LOS ( P < .001 ) and non-routine discharge ( P < .001 ).

Conclusion

Our study is the first to use the HFRS to assess the impact of frailty on patients with primary spinal tumors. We found that frailty was associated with prolonged LOS, non-routine discharge, and increased hospital costs.

Keywords

frailty intradural/cord tumor outcomes length of stay complications surgery

Introduction

Over the past decade, American healthcare spending has continued to rise, increasing the incentive for payors to identify mechanisms of reducing costs.^1-6 However, in the effort to minimize spending, it is also crucial to ensure the same quality of care is provided. In national payor-level datasets, proxies commonly used to indicate quality include hospital length of stay (LOS), postoperative complications, and non-routine discharge.^7-9 Optimization of these outcomes also appears to correlate with lower costs and higher-quality care in spinal surgery, where extended LOS and non-routine discharge have been identified as both drivers of high costs^7-9 and inferior surgical outcomes, including increased rates of complication and morality.^8,9 However, most of these investigations have been in those patients treated for degenerative pathologies vs spinal tumors. Intradural tumors of the spine comprise 40–45% of primary spinal neoplasms, commonly presenting with pain and progressive sensory or motor deficits.^10,11 Surgical resection is the gold standard for most intradural tumors,^12,13 but can be associated with high costs and rates of complication.¹⁰ Therefore, further investigation into patient risk-factors associated with inferior outcomes and increased resource utilization is warranted to better patient care.

One such factor that has garnered increased attention of late is frailty, a common condition among the elderly that is defined by decreased physical reserve and which may affect up to 30% of patients with primary spinal tumors.^14-16 Prior investigations of frailty in spine oncology have used validated measures such as the modified 5-item frailty index (mFI-5), modified 11-item frailty index (mFI-11), and metastatic spinal tumor frailty index (MSTFI).^17,18 However, most of these have been deployed in single-institution datasets and there is increased interest in studying frailty utilizing large, national databases.^19,20 One metric of frailty designed specifically for use in these datasets is the hospital frailty risk score (HFRS), which is calculated using 109 internationally standardized and easily accessible ICD-10 diagnostic codes for conditions overrepresented in frailty.^17,18,20 The HFRS has been validated as a reliable predictor of extended hospital LOS, intensive care unit (ICU) stays, and non-home discharge in patients undergoing surgery for a degenerative spinal condition.²¹ However, it has yet to be applied to patients with primary tumors of the spinal cord and meninges.

The aim of this study was to utilize the HFRS to investigate the impact of frailty on complication rates, LOS, discharge disposition, and total cost of hospital admission in patients undergoing surgical intervention for primary spinal intradural/cord tumors.

Methods

Data Source and Patient Population

The Healthcare Cost and Utilization Project’s Nationwide Inpatient Sample (NIS) database is a stratified discharge database representing 20% of all inpatient admissions from community hospitals in the United States. It is the largest all-payer healthcare database in the US, containing over 7 million hospital admissions (approximately 35 million hospitalizations, weighted) per year. A retrospective study was performed using years 2016–2018 of the NIS for all adult (≥ 18 years old) patients undergoing surgical intervention for primary spinal intradural/cord tumor. Institutional Review Board (IRB) was deemed exempted due to the deidentification of patients in the NIS database.

Within the NIS, the International Classification of Diseases, Tenth Revision, Clinical Modification [ICD-10-CM] diagnosis and procedural coding system (PCS) was used to identify patients with spinal intradural/cord tumors, as well as their previously diagnosed comorbidities and the surgical interventions they underwent. The ICD-10-CM codes C72.0 “Malignant neoplasm of spinal cord”, D32.1 “Benign neoplasm of spinal meninges”, D33.4 “Benign neoplasm of spinal cord”, D43.4 “Neoplasm of uncertain behavior of spinal cord” were used to identify our patient population. Patients with ICD-10-CM codes corresponding to metastatic cancer, peripheral nerve sheath tumors and secondary neoplasms of the central nervous system, and patients who underwent excision of vertebrae were excluded from analysis. Identified patients were then cross-matched using ICD-10-CM codes to determine what surgical interventions the patient underwent.

Hospital Frailty Risk Score (HFRS)

Frailty was defined using the HFRS, a validated ICD-10 coding algorithm that was developed by Gilbert et al.²⁰ and assessed in subsequent work.²² The coding algorithm is derived from a cluster of frail individuals with a comprehensive 109 ICD-10 codes assigned to a number of points ranging from .1 to 7.1 correlating to the strength of association predicting membership to the frail cluster.²⁰ These points are added together for a final frailty risk score.²⁰ All ICD-10 codes for the admission were entered into the calculation. In addition to presenting frailty as a continuous variable, we then dichotomized the frailty score into Non-Frail (HFRS<5) and Frail (HFRS>5).

Data Collection

Patient demographics such as age, sex, race, median household income, and insurance provider were all collected from the NIS database. Additional hospital characteristics such as size by bed volume, region (Northeast, South, Midwest, and West), and type (rural, urban teaching, and urban non-teaching) were also collected. Elixhauser comorbidities were used to evaluate incidence of congestive heart failure (CHF), cardiac arrhythmia, valvular disease, hypertension (HTN), pulmonary circulation disorders, peripheral vascular disorders, paralysis, other neurological disorders, chronic pulmonary disease, diabetes, hypothyroidism, renal failure, liver disease, peptic ulcer disease excluding bleeding, AIDS/HIV, lymphoma, metastatic cancers, solid tumor without metastasis, rheumatic disease, coagulopathy, obesity, weight loss, fluid and electrolyte disorders, anemia (blood loss or deficiency), alcohol or drug abuse, and psychoses. In addition, incidences of affective disorders, nicotine dependence, opioid dependence were assessed. Data on intraoperative variables such as procedural type (e.g., excision of the meninges/cord, release of spinal meninges/cord, cervical fusion, cervicothoracic, thoracic fusion, thoracolumbar fusion, lumbar fusion, and lumbosacral fusion), whether blood transfusion was required, and postoperative complications were also collected. Of note, HFRS utilizes only diagnostic coding; no procedural coding was identified in making the HFRS algorithm.

Data regarding postoperative complications for each patient was collected by indexing additional diagnoses from the NIS database. Complications included in the analysis were urinary tract infection (UTI), post-procedural fever, pressure ulcer, acute post-hemorrhagic anemia, displacement of internal fixation device of vertebrae, wound disruption, nervous system issues, acute deep vein thrombosis, post-procedural shock or infection, mechanical device complication, hematoma, post-procedural digestive ailments, and any anesthesia-related adverse event. In addition, postoperative outcome measures such as hospital LOS, total cost of hospital admission, and discharge disposition were also assessed between the 2 cohorts. Discharge disposition was classified as routine (patient went home), non-routine (patient sent to short-term hospital, skilled nursing facility, intermediate care facility, home with healthcare services), and other (leaving against medical advice, died in hospital, unknown destination). All-payer inpatient cost-to-charge ratios (CCR) were used to convert total hospital charge to total cost of hospital services.

Statistical Analysis

Discharge-level weights provided by the Agency for Healthcare Research and Quality were used to calculate national estimates. Parametric data was expressed as mean ± SD and compared via two-way independent t tests. Nonparametric data was expressed as median (interquartile range) and compared via the Kruskal–Wallis tests. Nominal data was compared with the χ² test. For our primary hypothesis, weighted univariate and multivariate logistic regressions were fitted with extended postoperative hospital LOS (as defined by LOS greater than the 75th percentile for the entire cohort), non-routine discharge, and increased hospital costs (as defined by total costs greater than the 75th percentile for the entire cohort) as the dependent variable. There were no patients with missing information on LOS. Patients with missing total cost variable were excluded from this portion of the analysis. Patients with “other” discharge were also excluded from this portion of the analysis to dichotomize routine vs non-routine discharge. Backward stepwise multivariate logistic regression analysis was used to select variables in the final model, using P ≤ .1 as entry and stay criteria. A P-value of less than .05 was determined to be statistically significant. Statistical analysis was performed using R Studio, Version 3.6.2, RStudio Inc, Boston, MA.

Results

Patient Demographics and Hospital Characteristics

Of the 5955 patients identified, 1260 (21.2%) were considered Frail, Table 1. While the Frail cohort was significantly older than the Non-Frail cohort (Non-Frail: 55.7 ± 16.4 years vs Frail: 63.3 ± 16.2 years, P <.001 ), there were no significant differences in household income quartile (P = .729), racial makeup (P = .489), or the proportion of female patients (P = .097) observed between the cohorts, Table 1. Healthcare provider differed significantly between the cohorts ( P < .001), with most Non-Frail patients being covered by private insurers (52.9%) and most Frail patients being covered by Medicare (54.0%), Table 1. With respect to treating hospital characteristics, a significantly greater proportion of Non-Frail patients were treated at urban teaching hospitals compared to Frail patients (Non-Frail: 92.0% vs Frail: 88.5%, P = .013 ), Table 1. However, there no significant differences observed between the cohorts with respect to hospital bed size (P = .232) or hospital region (P = .116), Table 1.

Table 1.

Patient Demographics and Hospital Characteristics.

Variables	Non-Frail (N = 4695)	Frail (N= 1260)	P-Value
Age (Years)
Mean ± SD	55.7 ± 16.4	63.3 ± 16.2	<.001
Female (%)	62.6	68.3	.097
Race (%)			.489
White	72.9	72.5
Black	8.1	10.2
Hispanic	8.2	5.9
Other	10.7	11.4
Income quartile (%)			.729
0–25th	20.0	18.0
26–50th	24.5	22.9
51–75th	25.4	25.7
76–100th	30.1	33.5
Healthcare coverage (%)			<.001
Medicare	31.2	54.0
Medicaid	9.9	10.3
Private insurance	52.9	32.1
Other	6.0	3.6
Hospital bed size (%)			.232
Small	9.4	6.3
Medium	21.0	19.0
Large	69.6	74.6
Hospital region (%)			.116
Northeast	20.7	21.8
Midwest	21.0	26.2
South	36.0	28.6
West	22.4	23.4
Hospital type (%)			.013
Rural	0.7	0.0
Urban non-teaching	7.2	11.5
Urban teaching	92.0	88.5

Patient Comorbidities

The Frail cohort had an HFRS that was nearly 7 points higher than the Non-Frail cohort (Non-Frail: 1.5 ± 1.4 vs Frail: 8.3 ± 3.0, P < .001 ), Table 2. When considering individual comorbidities, affective disorder (Non-Frail: 16.5% vs Frail: 23.8%, P = .008 ), congestive heart failure (Non-Frail: 1.2% vs Frail: 5.2%, P < .001 ), cardiac arrhythmias (Non-Frail: 8.3% vs Frail: 17.9%, P < .001 ), peripheral vascular disorders (Non-Frail: 1.8% vs Frail: 4.4%, P = .016 ), hypertension (Non-Frail: 41.7% vs Frail: 61.9%, P < .001 ), paralysis (Non-Frail: 9.5% vs Frail: 35.3%, P < .001 ), depression (Non-Frail: 9.6% vs Frail: 14.3%, P = .033 ), hypothyroidism (Non-Frail: 11.6% vs Frail: 18.3%, P = .005 ), renal failure (Non-Frail: 2.2% vs Frail: 9.5%, P < .001 ), coagulopathy (Non-Frail: 1.9% vs Frail: 5.2%, P = .005 ), weight loss (Non-Frail: .6% vs Frail: 4.4%, P < .001 ), fluid and electrolyte disorders (Non-Frail: 4.2% vs Frail: 27.0%, P < .001 ), deficiency anemias (Non-Frail: .6% vs Frail: 2.8%, P = .004 ), and other neurological disorders (Non-Frail: 2.8% vs Frail: 10.3%, P < .001 ) were all significantly more common in the Frail patients, Table 2.

Table 2.

Admission and Patient Comorbidities.

Variables (%)	Non-Frail (N = 4695)	Frail (N = 1260)	P-Value
Frailty Risk Score
Mean ± SD	1.5 ± 1.4	8.3 ± 3.0	<.001
Affective disorder	16.5	23.8	.008
Nicotine dependence	9.8	6.3	.092
Opioid dependence	0.3	N < 10^a	-
Congestive heart failure	1.2	5.2	<.001
Cardiac arrhythmias	8.3	17.9	<.001
Valvular disease	2.4	4.0	.209
Pulmonary circulation disorders	0.7	1.6	.230
Peripheral vascular disorders	1.8	4.4	.016
Hypertension, combined	41.7	61.9	<.001
Paralysis	9.5	35.3	<.001
Other neurological disorders	2.8	10.3	<.001
Depression	9.6	14.3	.033
Chronic pulmonary disease	12.4	15.9	.139
Diabetes, uncomplicated	8.1	11.5	.102
Hypothyroidism	11.6	18.3	.005
Renal failure	2.2	9.5	<.001
Liver disease	1.2	0.8	.608
Peptic ulcer disease excluding bleeding	0.3	1.2	.082
Rheumatoid arthritis/collagen vascular diseases	2.3	4.0	.160
Coagulopathy	1.9	5.2	.005
Obesity	15.8	20.2	.093
Weight loss	0.6	4.4	<.001
Fluid and electrolyte disorders	4.2	27.0	<.001
Blood loss anemia	N < 10^a	0.8	-
Deficiency anemias	0.6	2.8	.004

^aSignifies that the count number is <10 and cannot be reported.

Intraoperative Variables

Operative approaches were similar between the cohorts, Table 3. Additionally, rates of intraoperative cerebrospinal leak/dural tear (Non-Frail: 1.3% vs Frail: 2.0%, P = .402) and blood transfusions (Non-Frail: 1.2% vs Frail: 2.8%, P = .065) did not differ significantly between groups, Table 3.

Table 3.

Intraoperative Variables.

Variables (%)	Non-Frail (N = 4695)	Frail (N = 1260)	P-Value
Electrophysiological monitoring	31.2	30.2	.749
Procedure type
Excision of spinal meninges/cord	91.4	83.7	<.001
Release spinal meninges/cord	15.8	23.0	.006
Cervical fusion	4.0	9.5	.001
Cervicothoracic fusion	1.4	3.2	.047
Thoracic fusion	6.5	8.7	.210
Thoracolumbar fusion	1.1	1.2	.865
Lumbar fusion	1.7	2.8	.270
Lumbosacral fusion	0.2	1.2	.033
Transfusion
Blood	1.2	2.8	.065
Platelet	0.4	N < 10^a	-
Complications
Cerebrospinal leak or dural tear	1.3	2.0	.402

^aSignifies that the count number is <10 and cannot be reported.

Postoperative Complications

Compared to the Non-Frail cohort, the Frail cohort had a significantly higher rate of postoperative complications ( P < .001 ), with 39.3% of Frail patients and only 8.7% of Non-Frail patients experiencing ≥1 complication, Table 4. Within the Frail cohort, 32.1% of patients had 1 complication while 7.1% had > 1 complication, Table 4. For the Non-Frail cohort, 7.7% had 1 complication while only 1.1% had >1 complication, Table 4. Complications that were more common in Frail patients included UTI (Non-Frail: 1.1% vs Frail: 24.2%, P < .001 ), pressure ulcer (Non-Frail: .2% vs Frail: 4.4%, P < .001 ), acute post-hemorrhagic anemia (Non-Frail: 4.7% vs Frail: 8.7%, P = .017 ), post-procedural shock (Non-Frail: .0% vs Frail: .8%, P < .001 ), and acute deep vein thrombosis (Non-Frail: .6% vs Frail: 2.4%, P < .013 ), Table 4.

Table 4.

Postoperative Complications.

Variables (%)	Non-Frail (N = 4695)	Frail (N = 1260)	P-Value
Urinary tract infection	1.1	24.2	<.001
Post-procedural fever	0.4	1.2	.160
Pressure ulcer	0.2	4.4	<.001
Pulmonary embolism	0.4	N < 10^a	—
Acute post-hemorrhagic anemia	4.7	8.7	.017
Displacement of internal fixation device of vertebrae	0.0	0.0	—
Post-procedural shock	0.0	0.8	.007
Wound disruption	0.2	0.8	.158
Post-procedural infection	0.0	0.0	—
Mechanical device complication	0.0	0.0	—
Post-procedural hematoma	0.0	N < 10^a	—
Nervous system complication	1.0	0.8	.810
Acute deep vein thrombosis	0.6	2.4	.013
Post-procedural digestive	N < 10^a	0.8	—
Any complication	8.7	39.3	<.001
Number of complications			<.001
0	91.3	60.7
1	7.7	32.1
>1	1.1	7.1

^aSignifies that the count number is <10 and cannot be reported.

Hospital Outcomes

Compared to Non-Frail patients, Frail patients had longer mean LOS (Non-Frail: 4.9 ± 3.7 days vs Frail: 9.2 ± 6.7 days, P < .001 ), greater average hospitalization costs (Non-Frail: $25,187 ± 15 509 vs Frail: $40,163 ± 29 777, P < .001 ), and higher rates of non-routine discharge (Non-Frail: 31.0% vs Frail: 66.7%, P < .001 ), Table 5.

Table 5.

Postoperative Inpatient Outcomes.

Variables	Non-Frail (N = 4695)	Frail (N = 1260)	P-Value
Length of stay (days)
Mean ± SD	4.9 ± 3.7	9.2 ± 6.7	<.001
Median [IQR]	4 [3–6]	7 [5–11]	<.001
Total cost of admission ($)
Mean ± SD	25 187 ± 15 509	40 163 ± 29 777	<.001
Median [IQR]	21 599 [15 549–30 257]	31 715 [23 327–46 787]	<.001
Disposition (%)			<.001
Routine	68.8	32.5
Non-routine	31.0	66.7
Other	0.2	0.8

Multivariate Analysis for Extended LOS

On multivariate regression analysis, Frailty was found to be a significant independent predictor of prolonged LOS [OR: 6.24, CI (4.27, 9.10), P < .001 ], Table 6. While experiencing 1 complication was found to significantly increase the odds of extended LOS on multivariate analysis [OR: 2.20, CI (1.46, 3.31), P < .001 ], the risk of extended LOS was even greater among patients with >1 complication [OR: 3.99, CI (1.29, 12.41), P = .017 ], Table 6. Other factors found to associate with extended LOS on multivariate analysis included identifying as Black race [OR: 1.78, CI (1.08, 2.94), P = .024 ] and having Medicare coverage [OR: 2.17, CI (1.19, 3.97), P = .012 ], Table 6. Being of the 76-100^th income quartile was found on multivariate analysis to significantly reduce the odds of prolonged LOS [OR: .56, CI (.34, .90), P = .016 ], Table 6.

Table 6.

Logistic Multivariate Regression Analysis for Extended Length of Stay.

	Univariate Model	Multivariate Model	P - Value
Frailty	6.55 (4.77, 9.00)	6.24 (4.27, 9.10)	<.001
Age	1.01 (.99, 1.01)	1.00 (.98, 1.01)	.471
Female sex	1.14 (.84, 1.53)	1.03 (.73, 1.46)	.846
Race
White	Reference
Black	2.02 (1.28, 3.21)	1.78 (1.08, 2.94)	.024
Hispanic	1.11 (.64, 1.91)	.98 (.52, 1.84)	.945
Other	1.26 (.83, 1.93)	1.18 (.72, 1.93)	.504
Income quartile
0–25^th	Reference
26–50^th	.62 (.41, .95)	.70 (.43, 1.15)	.159
51–75^th	.72 (.48, 1.09)	.80 (.49, 1.29)	.356
76–100^th	.55 (.36, .82)	.56 (.34, .90)	.018
Healthcare coverage
Medicare	Reference
Medicaid	1.73 (1.12, 2.67)	2.17 (1.19, 3.97)	.012
Private insurance	.52 (.38, .70)	.79 (.50, 1.26)	.325
Other	1.31 (.73, 2.33)	1.85 (.87, 3.94)	.113
Hospital bed size
Small	Reference
Medium	1.53 (.81, 2.91)	Removed
Large	2.01 (1.14, 3.57)	Removed
Number of complications
0	Reference
1	3.61 (2.53, 5.16)	2.20 (1.46, 3.31)	<.001
>1	9.29 (3.99, 21.63)	3.99 (1.29, 12.41)	.017

Multivariate Analysis for Non-Routine Discharge

On multivariate regression analysis, Frailty was found to be a significant independent predictor of non-routine discharge disposition [OR: 2.99, CI (2.14, 4.18), P < .001 ], Table 7. Further, experiencing 1 [OR: 2.01, CI (1.34, 3.03), P < .001 ] or >1 complication [OR: 3.25, CI (1.32, 8.01), P = .010 ] was found to significantly increase the odds of undergoing non-routine discharge, Table 7. Other factors that independently predicted non-routine discharge included age [OR: 1.02 per year, CI (1.00, 1.03), P < .019 ] and identifying as Black [OR: 2.30, CI (1.38, 3.81), P = .001 ], Table 7. Factors associated with significantly decreased odds of non-routine discharge disposition on multivariate analysis included being in the 51-75^th [OR: .55, CI (.37, .83), P = .004 ] or 76–100^th income quartiles [OR: .62, CI (.41, .94), P = .023 ], having private insurance [OR: .46, CI (.31, .67), P <.001 ] or other (not Medicare or Medicaid) insurance [OR: .34, CI (.17, .71), P = .004 ], obtaining treatment at a western hospital [OR: .51, CI (.35, .76), P < .001 ], and having intraoperative electrophysiological monitoring [OR: .64, CI (.48, .85), P = .002 ], Table 7.

Table 7.

Logistic Multivariate Regression Analysis for Non-Routine Discharge Disposition.

	Univariate Model	Multivariate Model	P - Value
Frailty	4.55 (3.44, 6.02)	2.99 (2.14, 4.18)	<.001
Age	1.03 (1.02, 1.04)	1.02 (1.00, 1.03)	.019
Female sex	1.28 (1.00, 1.64)	.96 (.72, 1.27)	.770
Race
White	REFERENCE
Black	1.80 (1.15, 2.82)	2.30 (1.38, 3.81)	.001
Hispanic	.74 (.45, 1.23)	.87 (.49, 1.54)	.634
Other	1.05 (.71, 1.55)	1.20 (.80, 1.82)	.378
Income quartile
0–25^th	REFERENCE
26–50^th	.79 (.55, 1.14)	.77 (.50, 1.18)	.233
51–75^th	.65 (.46, .92)	.55 (.37, .83)	.004
76–100^th	.71 (.51, 1.01)	.62 (.41, .94)	.023
Healthcare coverage
Medicare	REFERENCE
Medicaid	.49 (.32, .75)	.60 (.34, 1.06)	.081
Private insurance	.30 (.23, .40)	.46 (.31, .67)	<.001
Other	.26 (.14, .47)	.34 (.17, .71)	.004
Hospital region
Northeast	REFERENCE
Midwest	.94 (.66, 1.34)	.75 (.49, 1.13)	.171
South	.65 (.47, .90)	.51 (.35, .76)	<.001
West	.92 (.63, 1.32)	.82 (.54, 1.25)	.353
Electrophysiological monitoring	.70 (.55, .90)	.64 (.48, .85)	.002
Number of complications
0	REFERENCE
1	2.97 (2.09, 4.21)	2.01 (1.34, 3.03)	<.001
>1	5.75 (2.38, 13.86)	3.25 (1.32, 8.01)	.010

Multivariate Analysis for Increased Cost

Although frailty was a significant independent predictor of increased cost on univariate analysis, it was not found to predict increased cost on multivariate regression analysis [OR: 1.37, CI (.89, 2.12), P = .152], Table 8. Furthermore, while experiencing 1 complication was found to significantly predict increased cost [OR: 1.71, CI (1.00, 2.90), P = .049 ], having >1 complication was not [OR: 2.21, CI (.70, 6.19), P = .132 ], Table 8. Other factors that independently predicted increased cost included having an extended LOS [OR: 1.36, CI (1.28, 1.44), P <.001 ], being of the 75–100th income quartile [OR: 2.33, CI (1.40, 3.90), P = .001 ], receiving treatment at a western hospital [OR: 3.04, CI (1.70, 5.42), P <.001 ], and getting intraoperative electrophysiological monitoring [OR: 2.00, CI (1.42, 2.80), P <.001 ], Table 8. Advanced age [OR: .98 per year, CI (.97, .99), P <.001 ] and being treated at a medium-sized hospital [OR: .39, CI (.20, .78), P =.008 ] were found to significantly predict a lower odds of hospital costs >75th percentile, Table 8.

Table 8.

Logistic Multivariate Regression Analysis for Increased Cost.

	Univariate Model	Multivariate Model	P - Value
Frailty	3.47 (2.57, 4.69)	1.37 (.89, 2.12)	.152
Age	.99 (.99, 1.00)	.98 (.97, .99)	<.001
Female sex	.92 (.70, 1.20)	.87 (.62, 1.22)	.430
Race
White	REFERENCE
Black	1.21 (.75, 1.93)	Removed
Hispanic	1.42 (.88, 2.31)	Removed
Other	1.58 (1.07, 2.34)	Removed
Income quartile
0–25^th	REFERENCE
26–50^th	.79 (.51, 1.22)	.86 (.48, 1.54)	.619
51–75^th	1.29 (.86, 1.93)	1.48 (.85, 2.56)	.167
76–100^th	1.58 (1.07, 2.32)	2.33 (1.40, 3.90)	.001
Hospital bed size
Small	REFERENCE
Medium	.53 (.28, .98)	.39 (.20, .78)	.008
Large	.96 (.55, 1.69)	.70 (.39, 1.26)	.238
Hospital region
Northeast	REFERENCE
Midwest	.84 (.50, 1.42)	.87 (.47, 1.63)	.670
South	.54 (.33, .87)	.58 (.32, 1.05)	.074
West	2.24 (1.38, 3.63)	3.04 (1.70, 5.42)	<.001
Electrophysiological monitoring	1.65 (1.22, 2.24)	2.00 (1.42, 2.80)	<.001
Length of stay	1.33 (1.26, 1.39)	1.36 (1.28, 1.44)	<.001
Number of complications
0	REFERENCE
1	2.87 (2.00, 4.11)	1.71 (1.00, 2.90)	.049
>1	7.92 (3.45, 18.19)	2.21 (.79, 6.19)	.132

Discussion

In this retrospective, national database study of 5955 patients undergoing surgery for a primary spinal intradural/cord tumor, we found on univariate analysis that frailty was associated with longer LOS, higher rates of non-routine discharges, and increased total costs. Additionally, on multivariate analysis, frailty was found to be an independent predictor of extended LOS and non-routine discharge.

Given the association of frailty with increased vulnerability to adverse healthcare outcomes,^14,15 several groups have examined the prevalence of frailty within the primary spinal intradural/cord tumor population. In a retrospective study of 1589 patients from the NIS database surgically treated for primary spinal tumors between 2002 and 2011, Ahmed et al.¹⁶ found using the 9-item spinal tumor frailty index (STFI) that 20.2% of patients were considered mildly frail (STFI = 1), 6.0% were moderately frail (STFI = 2), and 2.2% were severely frail (STFI ≥ 3).¹⁶ In another NIS database study of 18 297 patients surgical treated for benign intradural tumors between 2002 and 2011, Kalakoti et al. showed that many patients presented with frailty-associated comorbidities such as chronic obstructive pulmonary disorder (18.6% of patients), congestive heart failure (2.9%), and hypertension (41.0%).²⁵ Similarly, in a prospective, single-institution study of 38 patients undergoing surgery for primary intradural extramedullary tumors, Chotai et al.²⁶ found that frailty-related comorbidities were common: 15% of patients had diabetes, 43% had hypertension, 30% had coronary artery disease, and 5% had a prior myocardial infarction or congestive heart failure.²⁶ Additionally, the authors showed that 68% of patients had an American Society of Anesthesiologists (ASA) grade of 3 or 4 and 32% had an ASA grade of 2, indicating that the overwhelming majority of patients were living with mild or severe systemic diseases which likely contributes to frailty.²⁶ In our study, 21.2% of patients undergoing surgery for a primary spinal intradural/cord tumor were considered to be frail by HFRS criteria. Moreover, as the patient population ages and life expectancies increase, the prevalence of frailty will continue to rise. Therefore, continued efforts to identify and better characterize frailty will allow for more targeted preoperative planning and risk stratification.

While the prevalence of frailty continues to rise among patients operated for primary tumors of the spinal cord and meninges, there have been few attempts to characterize the effect of frailty and frailty-associated factors on postoperative complication rates. In the aforementioned study by Ahmed et al., the authors found the odds of experiencing a postoperative complication significantly increased as frailty score progressed from mild (OR: 3.83), to moderate (OR: 6.80), and to severe (OR: 13.05).¹⁶ Similarly, in a retrospective NIS database study of 19017 patients who underwent surgery for a primary tumor of the spine, Patil et al. showed on multivariate analysis that the odds of experiencing a complication were increased for patients with 1 (OR: 1.4), 2 (OR: 1.6), and ≥3 (OR: 1.6) pre-existing comorbidities.²⁷ Similar to these prior studies, the present study also found frailty to be associated with increased complications. However, what remains unclear is whether the increased risk of complications in frail patients results from the frail phenotype as a whole, or if it is predominately driven by a subset of the comorbidities associated with frailty.

In addition to being associated with increased costs, prolonged hospitalizations following spine surgery are associated with decreased patient satisfaction and increased complication rates.⁹ Similar to examinations of frailty and postoperative complications, only a few studies have considered the impact of frailty on LOS following surgery for spine tumors. In the retrospective study of 2170 patients treated surgically for either a primary or secondary spinal tumor, Lakomkin et al. showed that frailty predicted extended LOS, with an OR of 14.91.²⁸ Interestingly, in the same study, the authors conducted a subgroup analysis that examined the impact of frailty on outcomes just within the 981 patients with a primary tumor and found an even larger impact on prolonged LOS (OR: 323.6 comparing mFI-11 value of 1 vs 0).²⁸ Similarly, in their study of patients treated for primary tumors, Ahmed et al.¹⁶ showed that mean LOS increased along with advancing frailty score, extending from 6.4 days for non-frail patients to 18.3 days for severely frail patients.¹⁶ With prolonged length of stay, there is an association with increasing costs. Using an alternative definition of frailty—the Charlson Comorbidity Index (CCI)—and 15545 patients from the NIS dataset treated surgically for primary intramedullary tumors, Sharma et al. demonstrated that a high CCI on presentation was predictive of an increased cost of hospital admission (OR: 1.95).²⁹ These results echoed those of Lakomkin et al.,²⁸ who found each additional point on the CCI score to be associated with a 23% increase in the odds of having a prolonged LOS. Analogously, our study determined that frailty independently predicted prolonged LOS and higher costs. Identifying how frailty impacts LOS, there is opportunities for creating enhanced recovery after surgery (ERAS) pathways.³⁰ While ERAS protocols are being better understood and implemented into practice, further investigations need to be explored for these observed discrepancies, as doing so may improve patient care and reduce healthcare costs.

The final metric considered here - non-routine discharge—is also associated with increased healthcare resource utilization and poorer patient outcomes. To our knowledge, no previous studies have directly examined the impact of frailty on non-routine discharge following surgery for primary spinal intradural/cord tumors. However, several have found frailty-associated factors to predict this endpoint. In the Sharma et al. study, patients with a high CCI ( ≥ 2) had a nearly 2-fold higher odds of experiencing a non-home discharge relative to those with low CCI (0-1).²⁹ Analogously, in a study of 18297 patients from the NIS database operated for benign intradural tumors, Kalakoti et al. showed the following frailty-associated comorbidities to independently predict unfavorable discharge: coronary artery disease (OR: 1.47), congestive heart failure (OR: 1.80), coagulopathy (OR: 1.83), and hypertension (OR: 1.23).²⁵ Moreover, in a retrospective, NSQIP database study of 1232 patients who underwent laminectomy for an intradural extramedullary spinal tumor between 2011 and 2014, Ahn et al. demonstrated on multivariate analysis that specific frailty-associated conditions such as an age of ≥65 years (OR: 2.73) and dependent functional status (OR: 4.30) were associated with non-routine discharge.³¹ Likewise, our study found frailty to independently predict non-routine discharge within this patient population. As previously mentioned though, further work is necessary to determine whether the risk is derived from the frail phenotype as a whole, or whether it is predominately driven by a subset of the considered comorbidities. Regardless, it appears that improved preoperative screening that focuses on identifying frail patients may allow for the tailoring of perioperative care pathways so as to reduce adverse events and expedite safe home discharge following surgery for a primary spinal tumor.

This study has several limitations inherent to all studies employing administrative datasets that may impact the generalizability of the findings. First, given that the data was collected by ICD codes, the data may contain coding and reporting biases, misclassified diagnoses and procedures, or incomplete data. Second, as the NIS only has information that pertains to a single inpatient hospital admission, we were unable to assess more long-term outcomes. Third, as the data was analyzed retrospectively, we are unable to assess the impact of potential confounding factors, such as variations in surgical approach and technique which may have significant implications on our results. Lastly, we are comparing outcomes based on frailty. Many different operating definitions have been employed for frailty in the medical literature and it is unclear if any single definition is optimal. Here we used the HFRS and so it is possible that the findings may not directly translate to clinical settings in which an alternate definition, for example, the mFI-11, is employed. Nevertheless, the similarity of the present findings to those in the literature suggests that frailty, more broadly, may be an important predictor of patients at risk for adverse outcomes. Additionally, herein we employed data from the largest inpatient healthcare database in the United States, reducing the risk of sampling bias present in many single-institution investigations.

Conclusion

This study was the first to use the Hospital Frailty Risk Score (HFRS) to assess the impact of frailty on outcomes following surgery for primary tumors of the spinal cord and meninges. We found that in this patient population, Frail patients—those with high HFRS scores—had longer hospitalizations, higher rates of non-routine discharge, and higher total hospitalization costs. Subsequent studies exploring these findings are necessary to further improve quality of patient care and reduce healthcare costs.

Footnotes

Declaration of conflicting interests

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

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Aladine A. Elsamadicy, MD

Zach Pennington, MD

Andrew Hersh

C. Rory Goodwin, MD, PhD

References

Adogwa

Lilly

Khalid

, et al.

Extended length of stay after lumbar spine surgery: Sick patients, postoperative complications, or practice style differences among hospitals and physicians?

World Neurosurgery. 2019;123:e734-e739.

Mercer

Singh

Kanzaria

. Reducing emergency department length of stay. JAMA. 2019;321:1402.

Jiang

Upfill-Brown

Dann

, et al. Association of hospital length of stay and complications with readmission after open pancreaticoduodenectomy. JAMA Surgery. 2019;154:88.

Gay

Hall

Markham

Bettenhausen

Doupnik

Berry

. Association of extending hospital length of stay with reduced pediatric hospital readmissions. JAMA Pediatr. 2019;173:186-188.

Burner

Waldo

McKusick

. National health expenditures projections through 2030. Health Care Financing Review. 1992;14:1-29.

Morgan

. USA healthcare annual spending estimated to rise by 5.8% on average through 2024. American health & drug benefits. 2015;8:272.

Stopa

Robertson

Karhade

, et al. Predicting nonroutine discharge after elective spine surgery: external validation of machine learning algorithms. J Neurosurg Spine. 2019;31:742-747.

Yadla

Ghobrial

Campbell

, et al. Identification of complications that have a significant effect on length of stay after spine surgery and predictive value of 90-day readmission rate. J Neurosurg Spine. 2015;23:807-811.

De la Garza-Ramos

Goodwin

Abu-Bonsrah

, et al. Prolonged length of stay after posterior surgery for cervical spondylotic myelopathy in patients over 65years of age. J Clin Neurosci. 2016;31:137-141.

10.

Elsamadicy

Koo

Reeves

, et al. Cost and health care resource utilization differences after spine surgery for bony spine versus primary intradural spine tumors. World Neurosurgery. 2021;151:e286-e298. doi:10.1016/j.wneu.2021.04.015.

11.

Mende

Krätzig

Mohme

Westphal

Eicker

. Keyhole approaches to intradural pathologies. Neurosurg Focus. 2017;43:E5.

12.

Abul-Kasim

Thurnher

McKeever

Sundgren

. Intradural spinal tumors: current classification and MRI features. Neuroradiology. 2008;50:301-314.

13.

Yahanda

Buchowski

Wegner

. Treatment, complications, and outcomes of metastatic disease of the spine: from Patchell to PROMIS. Ann Transl Med. 2019;7:216.

14.

Clegg

Young

Iliffe

Rikkert

Rockwood

. Frailty in elderly people. Lancet. 2013;381:752-762.

15.

van den Berg

MEL

Castellote

Mahillo-Fernandez

de Pedro-Cuesta

. Incidence of spinal cord injury worldwide: A systematic review. Neuroepidemiology. 2010;34:184-192.

16.

Ahmed

Goodwin

De la Garza-Ramos

, et al. Predicting short-term outcome after surgery for primary spinal tumors based on patient frailty. World Neurosurgery. 2017;108:393-398.

17.

Shebeshi

Dolja-Gore

Byles

. Validation of hospital frailty risk score to predict hospital use in older people: Evidence from the Australian longitudinal study on women's health. Arch Gerontol Geriatr. 2021;92:104282.

18.

McAlister

van Walraven

. External validation of the hospital frailty risk score and comparison with the hospital-patient one-year mortality risk score to predict outcomes in elderly hospitalised patients: a retrospective cohort study. BMJ Qual Saf. 2019;28:284-288.

19.

Theou

Brothers

Mitnitski

Rockwood

. Operationalization of frailty using eight commonly used scales and comparison of their ability to predict all-cause mortality. J Am Geriatr Soc. 2013;61:1537-1551.

20.

Gilbert

Neuburger

Kraindler

, et al. Development and validation of a hospital frailty risk score focusing on older people in acute care settings using electronic hospital records: an observational study. Lancet. 2018;391:1775-1782.

21.

Hannah

Neifert

Caridi

, et al. Utility of the hospital frailty risk score for predicting adverse outcomes in degenerative spine surgery. Neurosurgery. 87; 2020:1223-1230.

22.

Acosta

Goldberg

Amin

. Evaluating the impact of frailty on periprocedural adverse events and mortality among patients with GI bleeding. Gastrointest Endosc. 2021;94:517-525. doi:10.1016/j.gie.2021.03.021.

23.

Kalakoti

Missios

Menger

Kukreja

Konar

Nanda

. Association of risk factors with unfavorable outcomes after resection of adult benign intradural spine tumors and the effect of hospital volume on outcomes: an analysis of 18, 297 patients across 774 US hospitals using the national inpatient sample (2002-2011). Neurosurg Focus. 2015;39:E4.

24.

Chotai

Zuckerman

Parker

, et al. Healthcare resource utilization and patient-reported outcomes following elective surgery for intradural extramedullary spinal tumors; Neurosurgery. 81(4):613-619;2017.

25.

Patil

Lad

Boakye

. Complications and outcomes after spinal cord tumor resection in the United States from 1993 to 2002. Spinal Cord. 2008;46:375-379.

26.

Lakomkin

Zuckerman

Stannard

, et al. Preoperative risk stratification in spine tumor surgery: A comparison of the modified Charlson index, frailty index, and ASA score. Spine. 1976;44:E782-E787.

27.

Sharma

Sonig

Ambekar

Nanda

. Discharge dispositions, complications, and costs of hospitalization in spinal cord tumor surgery: analysis of data from the United States nationwide inpatient sample, 2003-2010. Journal of Neurosurgery: Spine. 2014;20:125-141.

28.

Pennington

Cottrill

Lubelski

Ehresman

Theodore

Sciubba

. Systematic review and meta-analysis of the clinical utility of enhanced recovery after surgery pathways in adult spine surgery. Journal of Neurosurgery: Spine. 2021;34:325-347.

29.

Ahn

Phan

Cheung

White

SJW

Kim

Cho

SK-W

. Predictors of discharge disposition following laminectomy for intradural extramedullary spinal tumors. World Neurosurgery. 2019;123:e427-e432.

Hospital Frailty Risk Score and Healthcare Resource Utilization After Surgery for Primary Spinal Intradural/Cord Tumors

Abstract

Objective

Methods

Results

Conclusion

Keywords

Introduction

Methods

Data Source and Patient Population

Hospital Frailty Risk Score (HFRS)

Data Collection

Statistical Analysis

Results

Patient Demographics and Hospital Characteristics

Patient Comorbidities

Intraoperative Variables

Postoperative Complications

Hospital Outcomes

Multivariate Analysis for Extended LOS

Multivariate Analysis for Non-Routine Discharge

Multivariate Analysis for Increased Cost

Discussion

Conclusion

Footnotes

Declaration of conflicting interests

Funding

ORCID iDs

References