Sage Journals: Discover world-class research

Abstract

Study Design

Retrospective Review of a National Database.

Introduction

By utilizing a national database, this study aims to quantify the predictors of 30-day mortality after odontoid fixation and guide appropriate management for patients in whom the choice between operative and non-operative management is unclear.

Methods

The American College of Surgeons National Surgical Quality Improvement Database was queried using Current Procedural Terminology (CPT) codes and International Classification of Disease (ICD) codes to identify patients 60 or older who underwent surgical fixation of an odontoid fracture from 2005 to 2020. Risk factors for mortality significant in univariate and subsequent multivariate analysis were used to develop a scoring system to predict post-operative mortality.

Results

608 patients were identified. Patients were split into a non-mortality 30 days post-op group, and into a mortality 30 days post-op group. The following risk factors were included in the scoring system: functional dependency, disseminated cancer, albumin less than 3.5, WBC count greater than 16 k, anterior surgical approach, and pre-op SIRS. Using a cutoff value of 2, the CAAD-16 score had a sensitivity and specificity of 82% and 81%, respectively. The ASA score, cutoff at 4, showed a sensitivity and specificity of 64% and 75% respectively.

Conclusions

This sample of 294 patients represents one of the largest samples of odontoid fracture fixation patients available in the literature and comes from a nationally representative database. We structure relevant risk factors into the CAADS-16 score, which has the potential to be a clinically relevant tool to prevent short-term postoperative mortality.

Keywords

cervical geriatric trauma fixation odontoid

Introduction

In the United States, the elderly population is the fastest growing demographic, estimated to double by 2050.¹ In patients over the age of 65, the prevalence of upper cervical spine fractures is between 2.4% and 4.7%.^2,3 A study has shown that the rate of odontoid fractures in the geriatric population is rising exponentially causing an increasing need for clear treatment algorithms.⁴ Odontoid fractures are one of the most common upper cervical spine fractures in elderly patients and the most common fracture of the C2 vertebra.⁵ They represent about 10%–20% of all cervical fractures, and of these fractures, 60% can be classified as Type II.^6,7 In addition, 62.0% to 68.9% of these cervical spine injuries occur due to low energy falls from a standing position.^8,9 Costs associated with inpatient management of odontoid fractures is estimated to be approximately $1.5 billion¹⁰; therefore, there is considerable importance in optimizing management strategies from a value-based perspective.

While type I odontoid fractures are generally treated non-operatively, the decision-making process for the optimal treatment strategy in type II is far more complex. Associated factors including additional injuries, medical comorbidities, non-union risk factors and even patient preference are part of the decision process.¹¹ Distinguishing between patients who would do well with surgical treatment without an unacceptably high risk of mortality from those patients who would not is imperative to achieving the best outcomes.

Although there have been several studies evaluating postoperative mortality following odontoid fractures in elderly patients, there is currently no risk stratification system for predicting mortality after operative treatment. By utilizing a national database, this study aims to quantify the predictors of 30-day mortality after odontoid fixation and guide appropriate management for patients in whom the choice between operative and non-operative management is unclear.

Methods

The ACS NSQIP database was queried for all patients undergoing surgical fixation for odontoid fractures from 2006 to 2020. The database contains prospectively collected demographic and preoperative information, as well as 30-day postoperative outcomes for randomly selected patients at participating sites. Trained clinical reviewers input all data, and routine data audits are performed to maintain accuracy. Furthermore, quality-overall disagreement rates were reported as 2% in 2020.¹² Patients with a diagnosis of odontoid fracture were queried using both International Classification of Diseases (ICD) 9th and 10th edition diagnosis codes in addition to Common Procedural Terminology (CPT) codes (Supplemental Table 1) to better ensure that patients with multiple secondary diagnoses or procedures were not mistakenly included. In addition, only patients over 60 years of age were included to more accurately reflect the population in which the mortality prediction tool would be most useful. Patients with any missing data elements were excluded from final multivariate analysis. In addition, any variables with less than 10 total observations were also excluded from multivariate analysis.

Potential risk factors for mortality based on previous literature and univariate analysis were selected and included in univariate analysis. Mortality was calculated using the variable “yrdeath.” Anemia was calculated using preoperative hematocrit, under 39% for men and 36% for women. Hypoalbuminemia or malnutrition was calculated using preoperative albumin less than 3.5 mg/dL.

Using R statistical software (R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/), all included cases were subject to descriptive statistics and univariate analysis. Shapiro-Wilk tests were conducted to assess normality of data. If data was normal, two-sample Student t-tests were utilized. If data was not normal, Mann-Whitney U test was utilized. Fisher’s exact test was used to compare categorial variables when categories had five or fewer occurrences, while chi-squared was utilized when categories had greater than five occurrences. Multicollinearity was estimated using the variance inflation factor (VIF) with a threshold of 5 ⁴⁵. All factors found to be significant (P < .05) on univariate analysis were assessed via bivariate analysis. All variable significant in bivariate analysis (P < .05) were included a multivariable logistic regression model to build a risk estimating equation (REE), which was used to calculate predicted probability of 30-day mortality. A receiver operating curve (ROC) was generated, and analysis conducted to calculate the area under the curve (AUC) using the trapezoid rule.¹³ The Hosmer-Lemeshow test was used to assess goodness of fit.

The CAADS-16 score was derived as a practical tool for patients who are in consideration for operative fixation for odontoid fractures. CAADS-16 is an acronym: cancer, albumin less than 3.5, anterior surgical approach, functional dependency, pre-op SIRS, and WBC greater than 16k. For reference, the current study compared the CAADS-16 score to the American Society of Anesthesiologists (ASA) physical status classification score in its ability to accurately predict 30-day mortality. Significance was set at P < .05 for all statistical analyses.

This study did was exempt for IRB approval due to it utilizing a deidentified national database. This study also did not require consenting patients due to utilizing an available database.

Results

Cohort Characteristics and Univariate Analysis

608 total patients were identified using a combination of CPT and ICD 9/10 codes. Within this cohort, 308 patients were male and 300 were female. The average age was 76.7 years (SD 8.2). Information on the full cohort’s demographics, various lab data, and complications are shown in Table 1. Table 1 also presents other characteristics like ASA classification and ventilator dependency for the full cohort.

Table 1.

Characteristics of the Full Cohort.

Categories	All Patients (N = 608) N (%) or Average (Std. Dev)
Approach
Anterior	608 (100%)
Posterior	155 (25.5%)
Specialty doing operation
Orthopedics	140 (23.0%)
Neurosurgery	468 (77.0%)
Gender
Male	300 (49.3%)
Female	308 (50.7%)
Age at surgery (yrs)	76.7 (8.2)
Height (cm)	166.5 (11.1)
Weight (kg)	73.7 (19.1)
BMI (kg/m2)	26.4 (5.5)
Diabetes mellitus with oral agents or insulin	89 (14.6%)
Current smoker within 1 year	77 (12.7%)
Operative time (minutes)	152.0 (75.0)
Hospital admission to OR (days)	2.0 (2.6)
Labs test
Sodium (mmol/L)	137.5 (3.8)
BUN (mg/dL)	18.6 (11.4)
Creatinine (mg/dL)	1.0 (.8)
Albumin (g/dL)	3.7 (.4)
Bilirubin (mg/dL)	.6 (20.6)
SGOT (units/L)	27.6 (20.6)
Alkaline phosphate (IU/L)	85.3 (39.5)
WBC (10⁹/L)	8.2 (3.0)
HCT (male) (%)	38.3 (5.9)
HCT (female) (%)	37.0 (4.7)
Platelet (10⁹/L)	223.6 (71.0)
PTT (seconds)	29.7 (5.5)
INR	1.1 (.2)
PT (seconds)	12.3 (2.5)
In-patient or out-patient
In-patient cases	603 (99.2%)
Out-patient cases	5 (.8%)
Comorbidities
Dyspnea with moderate exertion	45 (7.4%)
Dyspnea at rest	5 (.8%)
Hypertension requiring medication	440 (72.4%)
History of severe COPD	58 (9.5%)
Immunosuppressive therapy	27 (4.4%)
Malnourishment	15 (2.5%)
Dependency
Partially dependent	74 (12.2%)
Totally dependent	11 (1.8%)
Ventilator dependent	16 (2.6%)
Ascites	2 (.3%)
Heart failure (CHF) in 30 days before surgery	13 (2.1%)
History of acute renal failure	2 (.3%)
Currently on dialysis	9 (1.5%)
History of disseminated cancer	5 (.8%)
Open wound/wound infection	25 (4.1%)
Bleeding disorder	52 (8.6%)
Infection
SIRS within 48 hours prior to surgery	46 (7.6%)
Sepsis within 48 hours prior to surgery	3 (.5%)
Septic shock within 48 hours prior to surgery	2 (.3%)
Emergent surgery	54 (8.9%)
ASA class
ASA1	1 (.2%)
ASA2	75 (12.3%)
ASA3	397 (65.3%)
ASA4	132 (21.7%)
ASA5	3 (.5%)
Outcomes
30 Day post-operative mortality	36 (5.9%)
30 Day post-operative complications	214 (35.2%)
Hospital stay (days)	7.5 (6.7)

ASA Class is the American Society of Anesthesiologists Class. PTT is the Partial Thromboplastin Time. PT is the Prothrombin Time. BUN is Blood Urea Nitrogen Test. SGOT is serum glutamic-oxaloacetic transaminase. HCT is percenta. SIRS is Systemic Inflammatory Response Syndrome. COPD is Chronic Obstructive Pulmonary Disease. OR is Operating Room.

Table 1 was recreated, but now splitting the cohort between patients who did suffer death within the 30-day post-operative period and patients who did not suffer death within the 30-day post-operative period as presented in Table 2. There were 36 (5.9%) patients who died following odontoid fixation and 572 (94.1%) who did not. Univariate analysis was conducted between all applicable variables. P-values are indicated in Table 2.

Table 2.

Comparison of Patients Who Suffered Mortality and Those Who Did Not.

Categories [N (%) or Average (Std. Dev)]	Mortality Group (N = 36)	Non-mortality Group (N = 572)	P-Value
Approach			.190
Anterior	13 (36.1%)	142 (24.8%)
Posterior	23 (63.9%)	430 (75.2%)
Specialty doing operation			.621
Orthopedics	10 (27.8%)	130 (22.7%)
Neurosurgery	26 (72.2%)	442 (77.3%)
Gender			.347
Male	21 (58.3%)	279 (48.8%)
Female	15 (41.7%)	293 (51.2%)
Age at surgery (yrs)	81.64 (7.4)	76.36 (8.2)	<.001
Height (cm)	167.4 (10.8)	166.4 (11.1)	.570
Weight (kg)	71.7 (25.2)	73.8 (18.6)	.212
BMI (kg/m²)	25.1 (6.2)	26.5 (5.5)	.062
Diabetes mellitus with oral agents or insulin	3 (8.3%)	86 (15.0%)	.390
Current smoker within 1 year	2 (5.6%)	75 (13.1%)	.287
Operative time (minutes)	162.1 (78.4)	150.9 (74.8)	.436
Hospital admission to OR (days)	3.6 (3.5)	1.9 (2.5)	<.001
Labs test
Sodium (mmol/L)	137.9 (4.6)	137.45 (3.8)	.993
BUN (mg/dL)	23.0 (20.9)	18.3 (10.4)	.066
Creatinine (mg/dL)	1.0 (.8)	1.0 (.8)	.892
Albumin (g/dL)	3.2 (.7)	3.7 (.5)	<.001
Bilirubin (mg/dL)	.7 (.3)	.6 (.4)	.051
SGOT (units/L)	39.6 (37.7)	26.8 (18.7)	<.001
Alkaline phosphate (IU/L)	109.0 (75.2)	83.7 (35.5)	.426
WBC (10⁹/L)	10.12 (4.1)	8.07 (2.8)	.002
HCT (male) (%)	33.3 (5.6)	38.7 (5.7)	<.001
HCT (female) (%)	38.5 (3.1)	37.0 (4.8)	.077
Platelet (10⁹/L)	233.1 (86.4)	223.0 (70.0)	.535
PTT (seconds)	28.6 (4.8)	29.78 (5.6)	.172
INR	1.2 (.2)	1.1 (.2)	.002
PT (seconds)	13.8 (4.6)	12.1 (1.9)	.446
In-patient or out-patient			1.000
In-patient cases	36 (100%)	567 (99.1%)
Out-patient cases	0 (0%)	5 (.9%)
Comorbidities
Dyspnea with moderate exertion	4 (11.1%)	41 (7.2%)	.330
Dyspnea at rest	1 (2.8%)	4 (.7%)	.264
Hypertension requiring medication	27 (75.0%)	413 (72.2%)	.864
History of severe COPD	5 (13.9%)	53 (9.3%)	.375
Immunosuppressive therapy	0 (0%)	27 (4.7%)	.396
Malnourishment	1 (2.8%)	14 (2.5%)	.604
Dependency
Partially dependent	10 (27.8%)	64 (11.2%)	.007
Totally dependent	1 (2.8%)	10 (1.8%)	.492
Ventilator dependent	9 (25.0%)	7 (1.2%)	<.001
Ascites	0 (0%)	2 (.4%)	1.000
Heart failure (CHF) in 30 days before surgery	2 (5.6%)	11 (1.9%)	.177
History of acute renal failure	0 (0%)	5 (.9%)	1.000
Currently on dialysis	2 (5.6%)	7 (1.2%)	.094
History of disseminated cancer	2 (5.6%)	3 (.5%)	.030
Open wound/wound infection	5 (13.9%)	20 (3.5%)	.012
Bleeding disorder	9 (25.0%)	43 (7.5%)	<.001
Infection
SIRS within 48 hours prior to surgery	8 (22.2%)	38 (6.6%)	.002
Sepsis within 48 hours prior to surgery	1 (2.8%)	2 (.4%)	.168
Septic shock within 48 hours prior to surgery	2 (5.6%)	0 (.0%)	1.000
Emergent surgery	7 (19.4%)	47 (8.2%)	.046
ASA class
ASA1	0 (.0%)	1 (.2%)	1.000
ASA2	2 (5.6%)	73 (12.8%)	.295
ASA3	14 (38.9%)	383 (67.0%)	.001
ASA4	18 (50.0%)	114 (19.9%)	<.001
ASA5	2 (5.6%)	1 (.2%)	.010
Hospital stay (days)	13.9 (13.7)	7.05 (5.8)	<.001

P-Values in bold indicate P < .05. ASA Class is the American Society of Anesthesiologists Class. PTT is the Partial Thromboplastin Time. PT is the Prothrombin Time. BUN is Blood Urea Nitrogen Test. SGOT is serum glutamic-oxaloacetic transaminase. HCT is percenta. SIRS is Systemic Inflammatory Response Syndrome. COPD is Chronic Obstructive Pulmonary Disease. OR is Operating Room.

Multivariable Analysis and Risk Model Development

These were the following variables included in the multivariable regression: anterior vs posterior approach, specialty (orthopedic vs neurosurgery), sex, age greater than 75 years, BMI, albumin levels less than 3.5 g/dL, white blood cell count greater than 12000 per microliter, HCT levels less than 38% for males and less than 35% for females, INR greater than 1, hypertension, functional dependency, dialysis, cancer, bleeding disorders, presence of preop SIRS, and length of time from admission to surgery greater than 2 days. Patients that did not have complete data for any of the previously mentioned variables were excluded. This left 294 patients for the multivariate analysis.

Using a multivariate regression detailed in Table 3, we can see the following variables were significant: anterior vs posterior approach, albumin levels less than 3.5 g/dL, white blood cell counts greater than 16000 WBC per microliter, functional dependency, disseminated cancer, and pre-op SIRS. The AUC of the logistic regression model was .89 (Hosmer-Lemeshow; P = .386). Multicollinearity was assessed, and none of the included variables in the multivariate model exceeded a threshold of 5 (all values were greater than 1 and less than 2).

Table 3.

Bivariate and Multivariate Logistic Regression With Significant Variables From Univariate Analysis.

	Bivariate Analysis		Multivariate Analysis
Variable	Odds ratio	P-Value	Odds ratio	P-Value	CI
Anterior vs posterior approach posterior approach ref	3.6	.009	6.9	.001	(2.2, 24.1)
Male vs Female	1.4	.457
Female ref	1.4	.457
Age ≥75 years	1.1	.084
Age <75 years ref	1.1	.084
BMI ≥30	1.0	.801
BMI <30 ref	1.0	.801
Albumin ≥3.5	0.2	<.001	4.9	.004	(1.7, 15.2)
Albumin <3.5 ref	0.2	<.001	4.9	.004	(1.7, 15.2)
WBC ≥16	1.2	<.001	16.1	.003	(2.5, 100.7)
WBC <16 ref	1.2	<.001	16.1	.003	(2.5, 100.7)
HCT ≥38 (male) 35 (female)	0.9	.137
HCT <38 (male) 35 (female) ref	0.9	.137
SGOT ≥45	1.0	.059
SGOT <45 ref	1.0	.059
INR ≥1	1.8	.513
INR <1 ref	1.8	.513
Hypertension	1.2	.736
No hypertension ref	1.2	.736
Functional dependency	2.9	.042	5.0	.007	(1.5, 16.5)
Not functionally dependent ref	2.9	.042	5.0	.007	(1.5, 16.5)
History of cancer	14.5	.009	19.2	.011	(1.6, 180.6)
No history of cancer ref	14.5	.009	19.2	.011	(1.6, 180.6)
Bleeding disorder	5.0	.003	3.4	.056	(.9,11.4)
No bleeding disorder ref	5.0	.003	3.4	.056	(.9,11.4)
Pre-op SIRS	5.5	.001	4.6	.009	(1.4, 14.7)
No pre-op SIRS ref	5.5	.001	4.6	.009	(1.4, 14.7)

P-Values in bold indicate P < .05. Hosmer-Lemeshow test was not significant (P = .871) indicating good fitting of multivariate model.

Creation of CAADS-16 Score

Using the previously outlined risk factors, the CAADS-16 score was developed. A single point was assigned for the presence of each factor, with a range of possible scores from 0 to 6. In the current sample, 119 patients had a score of 0, 106 had a score of 1, 62 had a score of 2, and 7 patient had a score of 3. No patients in our sample had scores of 4 or 5 or 6. Risk of mortality increased with higher scores; for a score of 0, 1, 2, and 3, observed mortality rates were 1.7%, 1.9%, 21.0%, and 71.4%, respectively (Figure 1). The AUC using the CAADS-16 score was .84. Using a cutoff value of 2 as a positive result, the CAADS-16 score has a sensitivity of 82% and a specificity of 81% (Figure 2). This was the cut-off value with the greatest sum of sensitivity and specificity, which also maximizes both these values.

Figure 1.

Box and whisker plot for each available CAADS-16 score in our cohort 165 x 103 mm (400 x 400 DPI).

Figure 2.

ROC curves for CAADS-16 score and ASA classification. The AUC is displayed in the legend, and the cut off values are represented in the squares of the curve. The ASA classification was cut off as 4 and the CAADS-16 score was cutoff at 2, chosen as the optimal cut offs for sensitivity and specificity.

Doing a similar analysis on the ASA classification the following was found. In the current sample, 0 patients had an ASA score of 1, 27 had an ASA score of 2, 186 had an ASA score of 3, 79 patients had an ASA score of 4, and 2 patients had an ASA score of 5. Risk of mortality increased with higher scores; for a score of 2, 3, 4, and 5 observed mortality rates were 0%, 4.3%, 15.2%, and 100%, respectively. The AUC using the AUC score was .72. Using a cutoff value of 4 as a positive result, the ASA score has a sensitivity of 64% and a specificity of 75% (Figure 2).

Discussion

In our study, we created the CAADS-16 score, to help predict 30-day mortality after odontoid fracture fixation in elderly patients. This tool will help physicians guide appropriate management for patients where the choice between operative treatment and non-operative treatment remains unclear. The ASA classification is a risk classification that has seen widespread use. But the agreement between surgeon judgements and the ASA physical status classification is lacking, especially between different specialties.¹⁴

The CAADS-16 score demonstrated an AUC of .84. Additionally, it performed better than the ASA classification, a commonly used classification system, to assess preoperative risk, is more focused toward mortality, and does not rely on subjective clinical assessment for any of its parameters.

A necessary criterion for the implementation of a new system is its ease of use in clinical practice.¹⁵ Regarding its practicality, the CAADS-16 score is relatively simple to implement and can be recorded accurately with minimal training. The simplicity of the components, with values of one for each risk factor, allows for ease of use which is a clear advantage in clinical situations where time and efficiency are paramount. The most appropriate use of the CAADS-16 score is to identify patients at high risk for mortality before surgery, allowing management to be tailored accordingly if necessary. It also provides a quantitative description of 30-day mortality, which can also be used to counsel patients and give them better insight into their given course of treatment.

Disseminated cancer has been previously described as a significant risk factor for surgical morbidity and mortality in various types of surgery for odontoid fractures.^16,17 In one such paper, patients receiving surgery for malignancy resection with a previous diagnosis of disseminated malignancy had 30-day mortality rates ranging from 33.7% to 26.6%, with variation based on year of surgery.¹⁸ In another study, patients with cancer had increased odds of mortality and morbidity compared to those without, with OR 3.32 (2.87-3.85) for mortality, and 1.28 (1.19-1.38) for morbidity as defined by the presence of any 30-day medical complication, readmission, or unplanned reoperation.¹⁹

Similarly, hypoalbuminemia has been cited as an independent risk factor for morality in various surgical specialties.^20-25 Finally, leukocytosis has also been investigated as an independent risk factor for postoperative complications. In patients receiving surgery for Stanford Type A aortic dissections, patients with WBC above 11k had significantly higher rates of mortality compared to those who did not -20.9% vs 8.1% (P = .001).²⁵ Other studies have observed similar relationships between increasing WBC count and risk of in-hospital mortality among patients with acute MI undergoing coronary artery bypass grafting.²⁶ There is, however, considerable conflicting evidence regarding the prognostic value of elevated WBC and mortality.²⁷ For example, among approximately 11,000 cardiac surgery patients, elevated WBC above 11k was associated with increased risk of infectious complications, but not mortality.²⁸ Given the inconclusive nature of the relationship between leukocytosis and mortality in the literature, it should be noted that the current study used a higher threshold WBC value of 16k as opposed to the more typical 11k upper range on most WBC assays. This may explain why elevated WBC was found to be significant in the current results and may be a topic for further investigation in future studies.

Functional dependency has been frequently cited as a predictor of mortality as well.^29-31 It is likely that functional status is a surrogate for a patient’s overall frailty/fragility, which has also been assessed as a factor positively correlated with postoperative mortality.³² Functional dependency has also been associated with increased post-operative morbidity and mortality in patients undergoing surgery for hip fracture.³³ Although this factor may be more difficult to objectively assess than presence of cancer, decreased albumin,³⁴ or elevated WBC count, it nevertheless represents an important prognostic factor in the CAADS-16 score, and may still be obtained with relative ease through history taking methods either from the patient, or from a designated caregiver.

Finally, anterior surgical approach for odontoid fracture fixation was also identified as a significant risk factor for 30-day mortality and included in the CAADS-16 score. This has previously been previously demonstrated in the literature.³⁵ Rates in the literature for 30-day overall mortality following odontoid fracture fixation with any technique range from 4% to 19.2% comparable to our own overall mortality rate of 7.8%.^35-37 In addition, 30-day mortality for patients undergoing anterior odontoid fixation has been reported in a previous systematic review from White et al to be 7% based on 4 prior studies.^38-42 Our own observed rate was 15.6%, which was higher than previously reported - the current study also encompassed a larger range of years and a larger sample size (294 in the current study vs 134 in White et al). While the precise reason for our higher observed 30-day mortality with anterior compared to posterior fixation is not entirely clear from the present results, it is in-line with prior expert recommendations regarding odontoid fracture fixation in the elderly.^43,44 These reviews recommend posterior C1-2 fixation for unstable fractures, and hard cervical collar or cervicothoracic bracing for stable fracture patterns. The current findings support these recommendations.

This study has several limitations. While CPT codes provide a codified way to identify procedures, the use of CPT codes may vary by surgeon and may not be fully representative of the procedure. We attempted to mitigate this by combining with ICD 9 and ICD 10 codes to choose cases more accurately. Due to the nature of ICD 9 and 10 codes, it was not possible to capture exact fracture patterns for all years of the included study - in particular, ICD 9 codes do not differentiate between Type I, II and III fracture patterns. ICD 10 codes do make this distinction and were tailored accordingly to capture only Type II fractures. Although operative management for Type I and Type III odontoid fractures is relatively rare, the potential inclusion of patients with these fracture patterns may impact results. Additionally, as this was a study utilizing a national database, it is susceptible to any inherent coding errors or inaccuracies present in the database itself. Furthermore, our study is retrospective and therefore subject to potential biases and confounding factors. For example, albumin levels could have only been obtained for patients who the operating surgeon suspected to be at a higher risk. Finally, due to the relatively small sample size and difficulty in identifying relevant risk factors in other databases that lack direct laboratory values such as NSQIP, it was not practical to externally validate the CAADS-16 score. In the future, external validation would be an important step in ensuring the generalizability of the CAADS-16 score to the patient population of interest. In addition, there were only 36 cases of mortality in this cohort. Due to this, our model could be subject to overfitting due to the number of covariates we utilized. We attempted to minimize for this by filtering by significant univariate parameters, then significant bivariate parameters. A large population is necessary for validation of these results, which can be done as more years of the NSQIP database are available. This would work to narrow the confidence intervals of our predictor variables.

Despite these limitations, the current study provides valuable insight into the risk factors for short-term mortality following odontoid fracture fixation. To our knowledge, this sample of 294 patients represents one of the largest samples of odontoid fracture fixation patients available in the literature and comes from a nationally representative database. Finally, we not only present the relative risk factors, but also structure them into the CAADS-16 score, which has the potential to be a clinically relevant tool for spine surgeons to prevent short-term postoperative mortality.

Conclusion

The CAADS-16 score is a simple and effective tool that can be used to identify patients at high risk of 30-day mortality following odontoid fracture fixation and can aid in the decision between operative and non-operative modalities. We believe that this score has the potential to improve clinical decision-making and patient outcomes in this patient population. However, further studies are needed to confirm the validity and generalizability of our findings.

Supplemental Material

Supplemental Material - Predicting Mortality Following Odontoid Fracture Fixation in Elderly Patients: CAADS-16 Score

Supplemental Material for Predicting Mortality Following Odontoid Fracture Fixation in Elderly Patients: CAADS-16 Score by William ElNemer, Eric Solomon, Michael Raad, Amit Jain, and Sang Hun Lee in Global Spine Journal.

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.

Permission

No copyrighted material used. AJ; Globus Medical: Paid consultant. Stryker: Paid consultant. DePuy, A Johnson and Johnson Company: Paid consultant. SL; Medtronic: Paid consultant; Paid presenter or speaker.

ORCID iDs

William ElNemer

Amit Jain

Supplemental Material

Supplemental material for this article is available online.

References

Goodkind

Kowal

. An aging world: 2015. International Population Reports. 2016;95:16.

Bub

Blackmore

Mann

, et al. Cervical spine fractures in patients 65 years and older: a clinical prediction rule for blunt trauma. Radiology. 2005;234:143.

Ngo

Hoffman

Mower

. Cervical spine injury in the very elderly. Emerg Radiol. 2000;7:287.

Zusman

Ching

Hart

, et al. Incidence of second cervical vertebral fractures far surpassed the rate predicted by the changing age distribution and growth among elderly persons in the United States (2005–2008). Spine. 2013;38:752-756.

Chapman

Bransford

. Geriatric spine fractures: an emerging healthcare crisis. J Trauma. 2007;62:S61.

Pearson

Martin

Lindsey

Mirza

. C2 vertebral fractures in the medicare population: incidence, outcomes, and costs. J Bone Joint Surg Am. 2016;98(6):449-456. PMID: 26984912. doi:10.2106/JBJS.O.00468.

Ryan

Henderson

. The epidemiology of fractures and fracture-dislocations of the cervical spine. Injury. 1992;23(1):38-40. doi:10.1016/0020-1383(92)90123-a. PMID: 1541497.

Golob

Jr Claridge

Yowler

, et al. Isolated cervical spine fractures in the elderly: a deadly injury. J Trauma. 2008;64:311.

Sokolowski

Jackson

Haak

, et al. Acute mortality and complications of cervical spine injuries in the elderly at a single tertiary care center. J Spinal Disord Tech. 2007;20:352.

10.

Daniels

Arthur

Esmende

Vigneswaran

Palumbo

. Incidence and cost of treating axis fractures in the United States from 2000 to 2010. Spine. 2014;39(18):1498-1505. doi:10.1097/BRS.0000000000000417.

11.

Pal

Sell

Grevitt

. Type II odontoid fractures in the elderly: an evidence-based narrative review of management. Eur Spine J. 2011;20(2):195-204. doi:10.1007/s00586-010-1507-6.

12.

American College of Surgeons . ACS NSQIP Participant Use Data File. American College of SurgeonsAmerican College of Surgeons; 2021, November 1: Retrieved March 25, 2023, from https://www.facs.org/quality-programs/data-and-registries/acs-nsqip/participant-use-data-file/

13.

Metz

. Basic principles of ROC analysis. Semin Nucl Med. 1978;8(4):283-298. doi:10.1016/s0001-2998(78)80014-2.

14.

Knuf

Maani

Cummings

. Clinical agreement in the American Society of Anesthesiologists physical status classification. Perioperat Med. 2018 Jun 19;7:14.

15.

Ferguson

. The rationale for developing scoring systems for clinical practice. Thorac Surg Clin. 2007 Aug;17(3):343-351.

16.

Al-Refaie

Parsons

Habermann

, et al. Operative outcomes beyond 30-day mortality: colorectal cancer surgery in oldest old. Ann Surg. 2011;253(5):947-952. doi:10.1097/SLA.0b013e318214b2a.

17.

Tuli

Patel

, et al. Odontoid fractures in the elderly: a review. Cureus. 2020;12(7):e9372. doi:10.7759/cureus.9372.

18.

Bateni

Meyers

Bold

Canter

. Current perioperative outcomes for patients with disseminated cancer. J Surg Res. 2015;197(1):118-125. doi:10.1016/j.jss.2015.03.063.

19.

Mullen

Michaels

Mehaffey

, et al. Risk associated with complications and mortality after urgent surgery vs elective and emergency surgery: implications for defining “quality” and reporting outcomes for urgent surgery. JAMA Surg. 2017;152(8):768-774. doi:10.1001/jamasurg.2017.0918.

20.

Larsen

Liest

Hannani

Jørgensen

Sørensen

Jørgensen

. Preoperative hypoalbuminemia predicts early mortality following open abdominal surgery in patients above 60 Years of age. Scand J Surg SJS Off Organ Finn Surg Soc Scand Surg Soc. 2021;110(1):29-36. doi:10.1177/1457496919888598.

21.

Gibbs

Cull

Henderson

Daley

Hur

Khuri

. Preoperative serum albumin level as a predictor of operative mortality and morbidity: results from the national VA surgical risk study. Arch Surg. 1999;134(1):36-42. doi:10.1001/archsurg.134.1.36.

22.

Eisenstein

Parry

Ramamoorthy

. Preoperative malnutrition with mild hypoalbuminemia associated with postoperative mortality and morbidity of colorectal cancer: a propensity score matching study. Nutr J. 2019;18(1):33. doi:10.1186/s12937-019-0458-y.

23.

Radcliff

Curry

Patel

, et al.

What is the incidence and severity of dysphagia after anterior cervical surgery?

Clin Orthop Relat Res. 2011;469(3):658-665. doi:10.1007/s11999-010-1671-y.

24.

Grosso

Hwang

Mroz

Benzel

Steinmetz

. Relationship between degree of focal kyphosis correction and neurological outcomes for patients undergoing cervical deformity correction surgery. J Neurosurg Spine. 2013;19(2):154-159. doi:10.3171/2013.4.SPINE12705.

25.

Shi

Feng

Wang

Liu

Wei

. The elevated admission white blood cell count relates to adverse surgical outcome of acute Stanford type a aortic dissection. J Cardiothorac Surg. 2020;15(1):48. doi:10.1186/s13019-020-1078-5.

26.

Newall

Grayson

, et al. Preoperative white blood cell count is independently associated with higher perioperative cardiac enzyme release and increased 1-year mortality after coronary artery bypass grafting. Ann Thorac Surg. 2006;81(2):583-589. doi:10.1016/j.athoracsur.2005.08.051.

27.

Robinson

Poisson

Pauyo

Hsu

. Risk factors for nonunion following surgical stabilization of odontoid fractures in adults: a systematic review and meta-analysis. Spine J. 2020;20(3):407-415. doi:10.1016/j.spinee.2019.08.462.

28.

Mahmood

Knio

Mahmood

, et al. Preoperative asymptomatic leukocytosis and postoperative outcome in cardiac surgery patients. PLoS One. 2017;12(9):e0182118. doi:10.1371/journal.pone.0182118.

29.

Scarborough

Bennett

Englum

Pappas

Lagoo-Deenadayalan

. The impact of functional dependency on outcomes after complex general and vascular surgery. Ann Surg. 2015;261(3):432-437. doi:10.1097/SLA.0000000000000767.

30.

Scarborough

Bennett

Englum

Pappas

Lagoo-Deenadayalan

. The impact of functional dependency on outcomes after complex general and vascular surgery. Ann Surg. 2015;261(3):432-437. doi:10.1097/SLA.0000000000000767.

31.

Curtis

Hammad

Anis

Higuera

Little

Darwiche

. Dependent functional status is a risk factor for perioperative and postoperative complications after total hip arthroplasty. J Arthroplasty. 2019;34(7S):S348-S351. doi:10.1016/j.arth.2018.12.037.

32.

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.

33.

Kammerlander

Gosch

Kammerlander-Knauer

, et al. Long-term functional outcome in geriatric hip fracture patients. Arch Orthop Trauma Surg. 2011;131(10):1435-1444. doi:10.1007/s00402-011-1338-6.

34.

Spallone

Izzo

Patrignani

Di Girolamo

Savini

Marfia

. Surgical treatment of odontoid fractures: anterior, posterior or combined approaches? A review of literature. Eur Rev Med Pharmacol Sci. 2019;23(4):1834-1844. doi:10.26355/eurrev_201902_17114.

35.

Molinari

Khera

Gruhn

. Functional outcomes, morbidity, mortality, and fracture healing in 58 consecutive patients with geriatric odontoid fracture treated with cervical collar or posterior fusion. Global Spine J. 2013;3(1):21-31. doi:10.1055/s-0033-1337122.

36.

Longo

Gelfand

De la Garza Ramos

, et al. Perioperative complications and mortality following anterior odontoid screw fixation in elderly patients: a national database analysis. World Neurosurg. 2019;129:e776-e781. doi:10.1016/j.wneu.2019.06.022.

37.

Chapman

Smith

Kopjar

, et al. The AOSpine north America geriatric odontoid fracture mortality study. Spine. 2013;38(13):1098-1104. doi:10.1097/BRS.0b013e318286f0cf.

38.

White

Hashimoto

Norvell

Vaccaro

. Morbidity and mortality related to odontoid fracture surgery in the elderly population. Spine. 2010;35(9S):S146. doi:10.1097/BRS.0b013e3181d830a4.

39.

Andersson

Rodrigues

Olerud

. Odontoid fractures: high complication rate associated with anterior screw fixation in the elderly. Eur Spine J Off Publ Eur Spine Soc Eur Spinal Deform Soc Eur Sect Cerv Spine Res Soc. 2000;9(1):56-59. doi:10.1007/s005860050009.

40.

Kuntz

Mirza

Jarell

Chapman

Shaffrey

Newell

. Type II odontoid fractures in the elderly: early failure of nonsurgical treatment. Neurosurg Focus. 2000;8(6):e7. doi:10.3171/foc.2000.8.6.8.

41.

Omeis

Duggal

Rubano

, et al. Surgical treatment of C2 fractures in the elderly: a multicenter retrospective analysis. J Spinal Disord Tech. 2009;22(2):91-95. doi:10.1097/BSD.0b013e3181723d1b.

42.

Smith

Vaccaro

Maltenfort

, et al. Trends in surgical management for type II odontoid fracture: 20 years of experience at a regional spinal cord injury center. Orthopedics. 2008;31(7):650.

43.

Hsu

Anderson

. Odontoid fractures: update on management. JAAOS - J Am Acad Orthop Surg. 2010;18(7):383.

44.

Konieczny

Gstrein

Müller

. Treatment algorithm for dens fractures: non-halo immobilization, anterior screw fixation, or posterior transarticular C1-C2 fixation. JBJS. 2012;94(19):e144. doi:10.2106/JBJS.K.01616.

45.

Kim

. Multicollinearity and misleading statistical results. Korean J Anesthesiol. 2019;72:558-569. Epub 2019 Jul 15. PMID: 31304696. doi:10.4097/kja.19087.

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

0.01 MB