Introduction
Among older adults, odontoid fractures are among the most clinically consequential cervical spine injuries, accounting for approximately 10%–15% of all cervical fractures and occurring more frequently than any other cervical spine fracture type in those older than 65 years [1,2]. With population aging, their incidence and related health care burden continue to increase. In the United States, hospitalization rates for C2 fracture in adults older than 84 years rose more than threefold, from 3.18 to 9.77 per 10,000 persons per year between 2000 and 2010 [3]. Despite advances in operative and nonoperative management, outcomes in older adults remain highly variable, with substantial morbidity and mortality not fully accounted for by fracture pattern or treatment strategy alone [4,5]. Consequently, increasing attention has been directed toward patient-specific factors that may better reflect physiologic vulnerability in this population.
Frailty and sarcopenia have emerged as complementary yet distinct markers of reduced physiologic reserve that may help explain variability in spine surgery outcomes beyond age alone [6,7]. In adult spine surgery and thoracolumbar trauma, both have been linked to poorer postoperative outcomes. Recent trauma literature suggests that frailty may be a stronger predictor of mortality than sarcopenia and may act synergistically with low muscle mass in particularly vulnerable patients [8–10]. This question is particularly pertinent in odontoid fracture management, where frailty has been shown to predict higher mortality independent of treatment approach in older adults with type II injuries [11]. However, the role of sarcopenia in this context is unclear, partly due to lack of standardized methods for measuring cervical sarcopenia. Although C3-based muscle assessments have been validated as practical surrogates for whole-body skeletal muscle mass on cervical imaging, prior work suggests that cervical cross-sectional area (CSA) may capture distinct biologic characteristics based on the muscle group studied and may not exclusively represent muscle quality [12–14].
Accordingly, the utility of cervical sarcopenia in preoperative risk assessment after odontoid fracture fixation remains uncertain. It is unclear whether different C3-based muscle measurements reflect the same underlying biologic vulnerability, whether any single cervical sarcopenia metric predicts postoperative risk comparably to frailty, or whether combining sarcopenia and frailty better identifies high-risk patients. Therefore, this study aimed to (1) compare three cervical sarcopenia metrics with frailty as predictors of unfavorable postoperative outcomes following surgical fixation of odontoid fractures; (2) characterize the intercorrelation and patient overlap among these cervical sarcopenia measures; and (3) determine whether combining cervical sarcopenia and frailty improves the identification of patients at elevated risk for postoperative complications.
Materials and Methods
Study design
This retrospective cohort study was conducted at a single urban academic quaternary level I trauma center. The study was conducted following the Strengthening the Reporting of Observational Studies in Epidemiology guidelines [15] and was granted Institutional Review Board exemption (HP-00108743). Adult patients aged ≥18 years who underwent surgery for traumatic odontoid fractures between January 2015 and December 2023 were identified from the departmental surgical billing database using Current Procedural Terminology codes. All surgeries were performed by one of four fellowship-trained orthopedic spine surgeons. Patients were excluded if perioperative cervical computed tomography (CT) imaging was unavailable or if postoperative follow-up was shorter than 30 days, except in cases of mortality within that period.
Data source and collection
Demographic, clinical, surgical, and postoperative data were collected through manual review of the electronic medical record, including inpatient and outpatient documentation. Recorded variables included demographics, comorbidities, surgical characteristics, and postoperative outcomes. Charlson comorbidity index and American Society of Anesthesiologists classification were obtained from preoperative anesthesia records. Polytraumatic injuries were categorized by involved body region, including head, thoracic, abdominal, pelvic, and extremity injuries.
Sarcopenia measurement
Cervical sarcopenia was evaluated using perioperative cervical CT imaging. Axial CT images at the C3 vertebral level were analyzed within the Picture Archiving and Communication Software (PACS, eUnity; Client Outlook Inc., Waterloo, ON, Canada) to quantify CSAs of the C3 vertebral body and bilateral cervical muscle groups. This methodology was based on prior studies demonstrating that C3-level muscle measurements can serve as a practical surrogate for skeletal muscle mass on cervical imaging [12,13]. Bilateral CSA measurements were obtained for the sternocleidomastoid (SCM), prevertebral, and paraspinal muscle groups at the C3 level.
Vertebral body area (VBA)-based normalization was selected to provide an anatomic adjustment for local patient size using measurements obtained from the same axial CT image, as both cervical muscle CSA and C3 VBA were measured at the same level. This approach was intended to generate cervical region-specific morphometric ratios for cohort-relative risk stratification rather than diagnose sarcopenia using consensus whole-body criteria. Height-normalized skeletal muscle index was not used because the primary aim was to compare cervical muscle measurements obtained from routine trauma CT imaging at the anatomical region of interest.
Frailty measurement
Frailty was graded using the 5-factor modified frailty index (mFI-5) [7], which includes diabetes mellitus, congestive heart failure, hypertension requiring medication, chronic obstructive pulmonary disease or pneumonia, and preoperative dependent functional status. Hypertension status was verified through a review of preoperative medication reconciliation records. Antihypertensive therapy was identified by medication class, including angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, calcium channel blockers, beta-blockers, thiazide and loop diuretics, centrally acting agents, and direct vasodilators. The mFI-5 was derived for each patient using these verified preoperative comorbidity data.
Outcomes
Postoperative outcomes of interest included overall mortality, any complication, non-home discharge, extended length of stay, revision surgery, motor deficit, sensory deficit, urinary tract infection, pneumonia, deep vein thrombosis, pulmonary embolism, wound infection, cerebrospinal fluid leak, nonunion, and fusion failure. Extended length of stay was defined as hospitalization exceeding the cohort median of 6.5 days.
Group assignment
For each sarcopenia metric, values were stratified into quartiles relative to the study sample. Patients in the first quartile exhibited the highest ratio values, reflecting the greatest muscle bulk, whereas those in the fourth quartile had the lowest ratio values and were considered the most sarcopenic. The mFI-5 was analyzed as an ordinal variable ranging from 0 to 5. An mFI-5 threshold of ≥2 was used to identify patients with increased frailty burden, defined as the presence of at least two mFI-5 deficits. This threshold was selected because it is clinically interpretable, has been used in prior spine literature to identify frail or high-frailty patients, and minimizes sparse subgroup comparisons at higher mFI-5 values within this cohort [16,17].
Statistical analysis
Categorical outcomes were compared between threshold-defined groups using Pearson χ2 tests or Fisher exact tests when expected cell counts were small. Threshold analyses compared patients in the lowest quartile of each sarcopenia metric with the remainder of the cohort and patients with an mFI-5 ≥2 with those below this threshold. Spearman rank correlation was used to evaluate intercorrelation among the three cervical sarcopenia metrics, and overlap among the lowest-quartile sarcopenia groups was assessed descriptively. Mann-Whitney U tests were performed to compare the ability of each continuous or ordinal predictor to distinguish between patients with and without each outcome, with effect sizes reported as r.
An exploratory subgroup analysis was performed to evaluate outcomes in patients who were both in the lowest quartile of at least one of the three sarcopenia metrics and had an mFI-5 ≥2. A Combined Sarcopenia-Frailty Score was subsequently constructed as a hypothesis-generating cumulative burden metric. Each sarcopenia metric was assigned a quartile rank from 1 to 4, with 1 representing the least sarcopenic quartile and 4 representing the most sarcopenic quartile. The three quartile ranks were summed to generate a cervical sarcopenia burden component ranging from 3 to 12, and the mFI-5 score was then added directly to yield a total score ranging from 3 to 17, with higher values indicating greater combined physiologic vulnerability. The three cervical muscle metrics were summed because they demonstrated only modest intercorrelation and partially overlapping lowest-quartile groups, suggesting that they may capture related but distinct aspects of cervical muscle morphology. An unweighted ordinal approach was selected because the cohort size and number of adverse events were limited, making regression-derived weighting susceptible to overfitting. Receiver operating characteristic analysis was performed to evaluate the discriminatory performance of this combined score for the exploratory adverse event composite. Area under the curve (AUC) values with 95% bootstrap confidence intervals (CIs) based on 2,000 iterations were calculated, and the optimal score threshold was identified using the Youden index. All analyses were conducted in Python ver. 3.12 (Python Software Foundation, Wilmington, DE, USA) with SciPy ver. 1.14 (SciPy Community, Austin, TX, USA), with statistical significance defined as p<0.05.
Results
Population characteristics
Sixty-one patients met the inclusion criteria. The cohort had a mean age of 61.7±12.4 years, and 31 patients (50.8%) were female. Most patients were White (73.8%), followed by Black patients (16.4%). The mean body mass index was 27.3±7.2 kg/m2, and 13 patients (21.3%) were current smokers. Polytrauma was observed in 40 patients (65.6%), most commonly affecting the head (57.4%) and extremities (31.1%). One patient (1.6%) sustained a spinal cord injury. Patients underwent instrumentation across a mean of 1.6±1 spinal levels, and fusion procedures were performed in 56 patients (91.8%). The mean hospital length of stay was 9±10.8 days, and the mean follow-up duration was 17±21.1 months (Table 1).
Predictors of interest
The SCM/VBA values ranged from 0.94 to 5.64, with a median of 2.49. Prevertebral/VBA values ranged from 0.59 to 3.33, with a median of 1.60, while paraspinal/VBA values ranged from 7.81 to 29.89, with a median of 14.30. Quartile assignment distributed 15–16 patients per quartile for each sarcopenia metric. The mFI-5 distribution was 0 in 22 patients (36.1%), 1 in 23 (37.7%), 2 in 12 (19.7%), 3 in 3 (4.9%), and 4 in 1 (1.6%), with a median score of 1. Overall, 16 patients (26.2%) demonstrated an mFI-5 score ≥2 (Table 2).
Intercorrelation of sarcopenia metrics
The 3 cervical sarcopenia metrics demonstrated only modest intercorrelation. Spearman correlation coefficients were 0.440 for SCM/VBA versus prevertebral/VBA (p<0.001), 0.468 for SCM/VBA versus paraspinal/VBA (p<0.001), and 0.335 for prevertebral/VBA versus paraspinal/VBA (p=0.008) (Table 3). Overlap between the most sarcopenic quartile groups ranged from 46.7% to 53.3%, suggesting that each metric identified only a partially overlapping subset of patients classified as sarcopenic (Table 4).
Threshold analysis
In threshold-based analyses, neither individual sarcopenia thresholds—defined by the most sarcopenic quartile (Q4)—nor an mFI-5 threshold of ≥2 was significantly associated with any assessed postoperative outcome (Table 5). The overall mortality rate was 3.3% (2/61), with both deaths occurring beyond 90 days postoperatively. A nonsignificant trend toward higher pneumonia rates was observed among patients with an mFI-5 ≥2 (18.8% vs. 1.8%, p=0.052).
Combined low-sarcopenia/high-frailty subgroup
Eleven patients (18%) demonstrated both severe cervical sarcopenia, defined by placement in the lowest quartile of at least one cervical metric and an mFI-5 ≥2. Compared with the remaining 50 patients, this subgroup demonstrated significantly greater rates of mortality (2/11 [18.2%] vs. 0/50 [0%], p=0.030), urinary tract infection (4/11 [36.4%] vs. 5/50 [10%], p=0.047), pneumonia (3/11 [27.3%] vs. 1/50 [2.0%], p=0.016), and nonunion (2/11 [18.2%] vs. 0/50 [0%], p=0.030). This subgroup accounted for all deaths and nonunions in the cohort, along with three of the four pneumonia cases and four of the nine urinary tract infections (Table 6).
Comparison of predictor discrimination
When analyzed as continuous or ordinal variables, the predictors demonstrated limited ability to discriminate most postoperative outcomes. Effect sizes were consistently small for any complication, non-home discharge, prolonged hospital stay, and fusion failure (r=0.00–0.20). Among the sarcopenia measures, prevertebral/VBA showed the strongest discrimination for mortality (p=0.022, r=0.29), followed by mFI-5 (p=0.014, r=0.32). In contrast, SCM/VBA (p=0.113) and paraspinal/VBA (p=0.280) were not significantly associated with mortality status (Table 7).
Combined Sarcopenia-Frailty Score
The Combined Sarcopenia-Frailty Score ranged from 4 to 15, with a median value of 8. In exploratory receiver operating characteristic analysis for the composite outcome of mortality, urinary tract infection, pneumonia, or nonunion—comprising 11 unique patients with at least one composite event—the score yielded an AUC of 0.73 (95% CI, 0.52–0.90). The Youden-optimal cutoff was ≥9, corresponding to a sensitivity of 73% and a specificity of 68% (J=0.41). Overall, 24 patients (39.3%) met this threshold, with event rates of 33% among patients with scores ≥9 compared with 8% among those with scores <9 (Fig. 1).
Discussion
In the present cohort, neither individual cervical sarcopenia metrics nor the mFI-5 consistently predicted postoperative complications, non-home discharge, or prolonged hospitalization following odontoid fracture fixation. However, the modest intercorrelation among these measures should be interpreted cautiously. Because the analysis relied solely on CT-derived CSA measurements, it remains unclear whether the limited overlap among SCM, prevertebral, and paraspinal measurements reflects true regional biological heterogeneity, differences in local muscle degeneration, or the intrinsic limitations of CSA-based assessment alone. CSA measurements do not distinguish contractile muscle tissue from fatty infiltration or other qualitative degenerative changes, which may differ among cervical muscle groups. Thus, these metrics should be interpreted as noninterchangeable cervical morphometric parameters rather than definitive representations of distinct sarcopenia phenotypes. These findings suggest that isolated muscle size measurements and brief frailty indices may incompletely characterize postoperative risk in this population. This relatively more modest standalone signal is not entirely discordant with the broader spine-related literature, where the prognostic utility of sarcopenia and frailty has varied across surgical populations and appears influenced by clinical context, case acuity, and measures used to define these constructs [8,18]. For example, a 2024 meta-analysis including 24 studies and 243,453 patients reported that sarcopenia was associated with increased overall adverse events after spine surgery, although these relationships remained heterogeneous across studies [19]. By contrast, prior odontoid-specific frailty work and higher-acuity spine cohorts suggest that physiologic reserve remains clinically relevant but may not be adequately captured by a single cervical muscle metric or brief frailty assessment alone [9–11,20]. Collectively, these findings support the concept that postoperative vulnerability following odontoid fracture fixation is likely multifactorial and may be incompletely characterized when sarcopenia or frailty are assessed independently.
Not all cervical sarcopenia metrics appeared to reflect the same underlying risk profile in this cohort. The three C3-based measurements demonstrated modest correlation (rs=0.34–0.47), and their lowest-quartile groups overlapped by only 47%–53%, suggesting that SCM, prevertebral, and paraspinal measurements frequently identified different subsets of at-risk patients. Clinically, this suggests that cervical muscle loss in odontoid fracture patients may not occur uniformly and that different muscle groups may reflect distinct anatomic or biologic domains of vulnerability. This interpretation aligns with prior cervical imaging studies demonstrating that C3-based muscle assessment is a practical surrogate for systemic skeletal muscle mass, but that cervical muscle morphology remains heterogeneous across regions [12,13]. Other cervical studies have likewise revealed that muscle CSA alone does not fully reflect muscle quality or degeneration, particularly in the setting of fatty infiltration [14]. These findings suggest that no single cervical muscle group can be assumed to represent global cervical sarcopenia in this population.
The strongest prognostic signal in this study emerged when sarcopenia and frailty were evaluated in combination rather than separately. Although patients exhibiting both low cervical muscle mass on any metric and an mFI-5 ≥2 comprised a relatively small subgroup, they accounted for all the deaths and nonunions in the cohort and experienced substantially higher rates of major medical complications. A similar pattern was noted with the exploratory Combined Sarcopenia-Frailty Score, where a threshold of 9 or greater produced an AUC of 0.73 and identified a subgroup with a 33% rate of significant adverse events, compared with 8% below this threshold. Clinically, these findings suggest that postoperative risk after odontoid fracture fixation may be most evident when reduced muscle reserve and systemic frailty coexist, rather than when either is assessed independently. This interpretation is consistent with emerging spine trauma-related literature suggesting that frailty may outperform sarcopenia as an isolated mortality predictor, while an mFI-5 threshold of 2 or greater may synergistically interact with sarcopenia to identify patients with particularly limited physiologic reserve [10]. However, the Combined Sarcopenia-Frailty Score should be interpreted cautiously. Because the score incorporates quartile rankings from three cervical muscle groups and one mFI-5 score, it is inherently weighted toward imaging-based sarcopenia burden rather than being equally balanced between sarcopenia and frailty. This weighting was not intended to suggest that cervical muscle measurements are more clinically important than systemic frailty but rather represented an exploratory attempt to integrate multicompartment cervical muscle loss with a brief frailty assessment. In addition, because the three muscle measurements were only modestly intercorrelated, the score may incorporate partially overlapping information across metrics. Therefore, the Combined Sarcopenia-Frailty Score should be considered a hypothesis-generating risk-stratification tool that requires external validation and comparison with alternative weighting models before clinical implementation.
This study should be interpreted considering several limitations. First, the retrospective single-institution design introduces potential selection bias and limits generalizability, especially since it only includes surgically treated odontoid fractures with accessible perioperative cervical CT imaging. Second, the cohort size was modest, and several outcomes of interest were infrequent, limiting statistical power and increasing the risk of type II error when evaluating individual sarcopenia metrics or frailty in isolation. Third, cervical sarcopenia was measured using CT-based CSA normalized to the C3 VBA. While this method provides a practical same-image adjustment for local skeletal size on routine cervical trauma CT, it should be regarded as an exploratory cervical morphometric surrogate rather than a consensus diagnostic definition of sarcopenia. Major sarcopenia definitions incorporate muscle strength, physical performance, and height-normalized muscle mass indices; therefore, VBA-normalized CSA may limit comparison with established skeletal muscle index thresholds and may not fully account for variation in sex, ethnicity, or overall body habitus [21,22]. In addition, CSA-based measurements do not capture muscle quality, fatty infiltration, strength, or functional capacity [23,24]. This limitation is particularly relevant to the observed modest intercorrelation among the three cervical muscle metrics, as CSA alone cannot determine whether differences between muscle groups reflect true biological heterogeneity or unmeasured variation in muscle quality on CT. Future studies incorporating magnetic resonance imaging-based muscle quality assessment, CT attenuation, fatty infiltration grading, or functional measures may better clarify whether multicompartment cervical muscle assessment provides prognostic value beyond simple CSA measurement. Fourth, the sarcopenia quartiles and combined score threshold were derived from this cohort and should therefore be viewed as exploratory rather than definitive clinical cutoffs. The Combined Sarcopenia-Frailty Score is intrinsically weighted toward imaging-based sarcopenia burden, and the inclusion of partially correlated cervical muscle metrics may introduce overlapping information. Hence, this score should not be interpreted as a validated prediction tool without external validation and comparison with alternative normalization and weighting approaches. Finally, although the hypertension status for the mFI-5 was independently confirmed through a review of medication records to improve accuracy, frailty was still assessed using a brief administrative index rather than a more comprehensive multidomain frailty assessment. Larger multicenter studies are needed to corroborate these findings, determine whether cervical muscle quality adds incremental prognostic value, and determine whether combined sarcopenia-frailty screening can be implemented as a practical preoperative risk-stratification tool.
Conclusions
Standalone cervical sarcopenia metrics and mFI-5 were not independently associated with adverse postoperative outcomes after odontoid fracture fixation. However, because the three cervical sarcopenia measurements revealed partially distinct at-risk patients and had only a modest intercorrelation, they could not be used interchangeably. In contrast, the coexistence of low cervical muscle mass and frailty identified a subgroup of patients with elevated rates of mortality, urinary tract infection, pneumonia, and nonunion. An exploratory Combined Sarcopenia-Frailty Score threshold ≥9 may offer a framework for combining multicompartment cervical muscle burden with systemic frailty; however, its internally derived cutoff and imaging-weighted structure require external validation before clinical use.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
