Outcomes following anterior odontoid screw versus posterior arthrodesis for odontoid fractures: a systematic review and meta-analysis

OBJECTIVE Odontoid fractures can be managed surgically when indicated. The most common approaches are anterior dens screw (ADS) fixation and

are anterior dens screw (ADS) fixation or posterior C1-C2 arthrodesis (PA).ADS fixation has the advantage of preservation of the atlantoaxial motion, but is associated with high rates of postoperative dysphagia and thought to have lower union rates, especially in elderly patients.It requires an intact transverse ligament, reduced odontoid, favorable fracture line, and good alignment.In contrast, PA can be used when reduction of atlantoaxial subluxation or displaced fragments is required.The main disadvantages of this technique are the resultant loss of atlantoaxial motion, usually longer operative duration, increased postoperative pain, and prone positioning.
The objective of the present study was to systematically review all available literature and compare ADS fixation versus PA for odontoid fractures in terms of their safety profile-including reoperation rates and technical failures-as well as their efficacy profile, with the primary outcome being bony fusion at follow-up.

Methods
This systematic review and meta-analysis was per-formed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. 9

Search Strategy and Selection Criteria
Systematic searches were conducted in PubMed/MED-LINE, EMBASE, and Cochrane Central databases.The search algorithm used for PubMed was the following: ("odontoid" OR "dens" OR "odontoid process") AND fracture and (anterior OR "odontoid screw" OR "dens screw") AND (posterior OR "C1-C2" OR arthrodesis OR fixation OR transarticular OR interlaminar).
The search was conducted by two independent investigators (S.M., P.T.).Any disagreements or discrepancies were resolved by a third investigator (C.L.K.).The references of the included studies were also manually reviewed in order to identify further eligible articles.
A study was included in this meta-analysis if it fulfilled three predefined criteria: 1) randomized controlled trials or prospective or retrospective observational analyses comparing ADS fixation versus PA for fractures of the odontoid (including rigid and semirigid techniques); 2) comparative studies that report quantitative data on clinical outcomes of interest; and 3) studies published up to May 2022.Studies that did not compare the two surgical techniques or studies that did not report the primary outcome were excluded.

Data Extraction
Data extracted included the first author, title, date of publication, country of origin, patient number, demographics, age, type of fracture, conservative management, type of surgical technique, and follow-up duration.The primary outcome was fusion at follow-up.Fusion was defined as the presence of trabeculation across the fracture line on radiographs.Few studies used CT to evaluate bony fusion.Secondary outcomes were technical failure, reoperation, and all-cause mortality.
Risk of bias was assessed by two investigators (S.M., P.T.) using the Risk of Bias in Nonrandomized Studies-of Interventions (ROBINS-I) tool for observational studies by Cochrane. 10Funnel plots and Egger's regression test were used for assessment of publication bias when more than 10 studies were included to synthesize the outcome of interest.

Statistical Synthesis and Analysis
Odds ratios with the corresponding 95% confidence intervals were used for categorical outcomes assessed using random-effects model meta-analysis.Heterogeneity was assessed with the Higgins I 2 statistic. 11An I 2 > 50% indicated significant heterogeneity.Forest plots were used to graphically display the effect size in each study and the pooled estimates.Funnel plots and Egger's regression test were used for assessment of publication bias when at least 10 studies were included to synthesize the outcome of interest. 12A p value < 0.05 was considered significant.Stata 14.1 (StataCorp) was used as statistical software.

Literature Search and Characteristics of the Included Studies
The search strategy identified a total of 881 studies after duplicates were removed.After title and abstract screening, 38 studies underwent full-text evaluation.Twenty-two studies fulfilled the selection criteria and were included for quantitative analysis as shown in the PRISMA flow diagram (Fig. 1).The assessment of risk of bias is presented in Table 1.

Discussion
This systematic review and meta-analysis of compara-tive studies only, included 963 patients in total who underwent surgical treatment for odontoid fracture.Patients in the ADS group were associated with statistically significantly lower odds of fusion and higher odds of reoperation compared to PA.The rates of all-cause mortality and technical failure were similar between the two groups.In patients > 60 years old the same trend persisted although with a stronger effect size in terms of fusion at follow-up; OR of 0.46 for the entire patient cohort versus OR of 0.24 for patients > 60 years old only.
Odontoid fractures and especially type II fractures are associated with particularly high rates of nonunion, reported as high as 40% in the literature if managed conservatively with a halo vest or cervical collar. 33The goal of surgical stabilization is to increase the rates of bony fusion and can be performed either via an ADS fixation or PA approach.Surgery can be considered especially in patients > 50 years old-noting the risk of complications in elderly patients, type II fractures, dens displacement > 5 mm, and lack of acceptable reduction and alignment while in a cervical collar.Selection of the approach is not always interchangeable and depends on patient-specific factors, fracture characteristics, and surgeon's preference.ADS fixation is an osteosynthetic approach with the major benefit of preserving C1-C2 motion.However, it should be noted that comminuted fractures, cervicothoracic kyphosis, severe osteoporosis, transverse ligament rupture, late fractures, and a fracture line that is not anterosuperior to posterosuperior are considered to be contraindications to the anterior approach.In addition, obesity associated with unsuitable body habitus may have an inherently higher risk of postoperative dysphagia and aspiration pneumonia, which can subsequently increase length of stay, morbidity, and mortality.
With this meta-analysis of comparative studies only, we showed that the posterior approach was associated with increased likelihood of bony fusion at last followup, with the cumulative rates being 84.1% and 92.3% for the ADS and PA groups, respectively.Interestingly, in our subgroup analysis including patients > 60 years old only, the effect size was even stronger-OR 0.46 for the entire patient cohort versus OR 0.24 for patients > 60 years old only.The cumulative rates of bony fusion in this subgroup were 72.4% for the ADS and 89.9% for the PA groups.Considering the above findings, it would have been useful to perform a subgroup analysis for younger patients (i.e., age < 50 years); it is possible that the statistically significant differences in fusion rates would disappear given that there is a stronger effect size in elderly patients.Unfortunately, that was not statistically feasible with the given data from the included studies.Compared to prior meta-analyses, 8,34 our study has provided almost double the sample size and has provided further evidence that the PA is superior to ADS fixation in terms of fusion and reoperation rates.In addition, our study used a randomeffects model meta-analysis; given the heterogeneity of patients/surgical approaches, we believe that this is more appropriate than a fixed-effects model, which was used by prior studies.
The rates of reoperation were higher in the ADS compared to the PA group, both in the entire cohort and in the subgroup of patients > 60 years old only.The cumulative rates and effect sizes were similar when comparing outcomes for the entire cohort and for patients > 60 years old only (entire cohort: ADS 12.4%, PA 5.2%, OR 2.56; > 60 years old only: ADS 13.8%, PA 6.7%, OR 2.67).Even though the rates of technical failure were not statistically different in our analyses, the cumulative rates show a potentially important clinical difference.More specifically, in our cohort analysis 2.3% of the ADS and 1.1% of the PA groups experienced a technical failure.This trend persisted in the subgroup analysis of patients > 60 years old only (ADS 5%, PA 1.9%).It is likely that statistical power was too low to detect a statistically significant difference between the two groups.All-cause mortality rates were similar between the two groups in both our cohort and subgroup analysis.

Limitations
Our results should be interpreted in the context of several limitations.First, results stemming from analyses of observational studies are subject to their inherent risk of bias due to the nonrandomized design and nonblinded nature.Second, in the real-world setting, even though all studies included in this meta-analysis reported a direct head-to-head comparison of ADS fixation versus PA, the inherent selection bias due to patient-related factors, surgeons' preference, and fracture characteristics could not be adjusted for.Third, even though it appears that age might be a significant factor affecting the rates of fusion, it would be interesting to investigate whether in younger patients ADS fixation can perform equally well or if it is superior to PA in terms of fusion rates.In addition, there is increased heterogeneity in the surgical approaches used by the included studies especially in the PA group.Unfortunately, the included studies did not consistently report outcomes based on the different PA techniques; therefore, whether a specific dorsal approach technique is superior to the other cannot be investigated in the present study.The majority of patients in the PA group underwent C1-C2 arthrodesis, and a smaller percentage underwent other rigid (C1-C3, occipitocervical fusion) versus semirigid (cables/ wires) versus temporary fixation techniques.Including the semirigid techniques in the PA group could in theory decrease the rate of bony fusion and, therefore, if we were able to perform a subgroup analysis excluding these cases, the effect size of bony fusion favoring the PA group would only be stronger.Also, three of the studies reported included both acute and chronic fractures but did not provide outcomes based on the chronicity of the fracture, which precludes us from performing further subgroup analyses.Attempts to obtain additional information were unsuccessful.Last, the range for duration of follow-up across the studies and was rather limited in some.Future prospective studies or national registries are required to validate our results.These studies should provide patient-level data to allow for identification of potential confounders and significant factors that can affect outcomes, including reporting outcomes based on the specific PA technique used.

Conclusions
This systematic review and meta-analysis of ADS fixation versus PA for odontoid fractures included 963 patients in total.Patients in the ADS group were associated with statistically significantly lower odds of fusion and higher odds of reoperation compared to PA, and the former would therefore be the preferred approach.The rates of all-cause mortality and technical failure were similar between the two groups both in the entire cohort and in the subgroup analysis.In patients > 60 years old only, ADS fixation was still associated with lower odds of fusion, with an even stronger effect size.Reoperation rates were still significantly higher in the ADS versus PA group in patients > 60 years old.

FIG. 3 .
FIG. 3. Forest plot showing the comparison of ADS versus PA in terms of reoperation.Figure is available in color online only.

FIG. 4 .
FIG. 4. Forest plot showing the comparison of ADS versus PA in terms of technical failure.Figure is available in color online only.

TABLE 1 . Risk of bias assessment with the ROBINS-I tool in 22 studies of odontoid fracture
ROBINS-I = Risk of Bias in Nonrandomized Studies-of Interventions.

TABLE 2 . Baseline characteristics of the patients, surgical techniques, and fractures in 22 studies of odontoid fracture
reported age had the following ranges among the studies (ADS 30-81 years, PA 22-82.4years).Important baseline patient, surgical technique, and fracture characteristics are summarized in Table2.The range of the average followup across the included studies was 3-51 months postoperatively.

TABLE 2 . Baseline characteristics of the patients, surgical techniques, and fractures in 22 studies of odontoid fracture
FIG. 2. Forest plot showing the comparison of ADS versus PA in terms of fusion.Figure is available in color online only.Unauthenticated | Downloaded 10/19/23 01:43 PM UTC odds of reoperation compared to the PA group (ADS 12.4%; PA 5.2%; OR 2.56; 95% CI 1.50-4.35;I

TABLE 3 . Baseline characteristics of studies of odontoid fracture included in the subgroup analyses for patients ≥ 60 years old
FIG. 5. Forest plot showing the comparison of ADS versus PA in terms of fusion for patients > 60 years old.Figure is available in color online only.