Surgical versus nonsurgical management for type II

The Spine Journal 18 (2018) 1921–1933

Review Article

Surgical versus nonsurgical management for type II odontoid fractures in the elderly population: a systematic review Deep P. Sarode, MBChB (Hons), BMedSci (Hons)a,c, Andreas K. Demetriades, MBBCHIR, MPHIL, MRCS(Ed), MRCS(Eng), FRCSa,b,c,* a

College of Medicine and Veterinary Medicine, The University of Edinburgh, 47 Little France Crescent, Edinburgh EH16 4TJ, United Kingdom b Department of Clinical Neurosciences, Western General Hospital, Crewe Rd South, Edinburgh EH4 2XU, United Kingdom c Edinburgh Spinal Surgery Outcomes Study Group, Department of Clinical Neurosciences, Western General Hospital, Crewe Rd South, Edinburgh EH4 2XU, United Kingdom Received 30 November 2017; revised 25 March 2018; accepted 16 May 2018

Abstract

BACKGROUND: Odontoid process fractures, of which type II constitute the majority, are an increasingly important cause of morbidity and mortality in the elderly population. The incidence of geriatric type II fractures is steadily increasing in line with the aging population. However, the decision between surgical and non-surgical intervention for type II fractures in the elderly remains controversial. PURPOSE: The present study aims to synthesize the current published literature comparing outcomes following surgical and non-surgical interventions for type II odontoid fractures in the elderly population (≥65 years old). STUDY DESIGN/SETTING: Systematic review and meta-analysis were performed. METHODS: A systematic search of MEDLINE, MEDLINE In-Progress & Other Non-Indexed Citations, Embase, and Cochrane Central Register of Controlled Trials (CENTRAL) was performed to identify available evidence in English language. Studies with extractable data for all type II odontoid fractures in participants aged 65 years or older and which compared surgical and non-surgical intervention were included. Methodological quality was assessed using the Downs & Black checklist. Primary outcomes were mortality at short-term follow-up (≤3 months), mortality at long-term follow-up (predetermined study endpoint or mean follow-up length), and radiological union rate. Funding was provided by The University of Edinburgh for travel expenses to present this paper at the Society of British Neurological Sciences 2016 Conference ($170). RESULTS: Twelve studies (n=1,098), all non-randomized, met eligibility criteria. Methodological quality was particularly poor in the confounding, bias, and power domains of assessment. Substantial methodological and statistical heterogeneity allowed only a narrative synthesis of the primary outcomes. Overall, data on mortality at short-term follow-up appeared to favor neither surgical nor non-surgical intervention. A small favorable outcome in surgically managed patients over nonsurgically managed patients in terms of mortality at long-term follow-up was not proven conclusive because of considerable heterogeneity in study methodologies. Inadequate reporting of the time point of union assessment introduced the potential for significant intra- and interstudy heterogeneity and precluded assessment of union rates. CONCLUSIONS: Evidence on this controversial topic is sparse, markedly heterogeneous, and of poor quality. Well-designed prospective trials adhering to guidance published by the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) initiative are required to inform clinical practice on this contentious but growing issue. Future randomized controlled trials should

FDA device/drug status: Not applicable. Author disclosures: DPS: Support for travel to meetings for the study or other purposes: The University of Edinburgh (A), pertaining to the submitted work. AKD: Nothing to disclose. The disclosure key can be found on the Table of Contents and at www.TheSpineJournalOnline.com. https://doi.org/10.1016/j.spinee.2018.05.017 1529-9430/© 2018 Elsevier Inc. All rights reserved.

* Corresponding author. Department of Clinical Neurosciences, Western General Hospital, Crewe Rd South, Edinburgh EH4 2XU, United Kingdom. Tel.: +441315372110. E-mail address: [email protected] (A.K. Demetriades)

1922

D.P. Sarode and A.K. Demetriades / The Spine Journal 18 (2018) 1921–1933

include an assessment of frailty and medical comorbidities with suitable patients subsequently randomized to surgical or non-surgical treatment. © 2018 Elsevier Inc. All rights reserved. Keywords:

Elderly; Non-surgical; Odontoid fracture; Surgical; Systematic review; Type II

Introduction Odontoid fractures Odontoid fractures constitute the most common cervical spine fracture in the population aged over 70 years old [1]. These fractures are associated with substantial, yet not fully characterized, morbidity and mortality in the elderly population, which have been proposed to be almost equivalent to that of hip fractures [2]. In addition, these fractures, which are the most common types of C2 fracture, pose a considerable socioeconomic burden on health-care services [3]. The Anderson and D’Alonzo classification is used to characterize these fractures based on the location of the fracture line in relation to the odontoid process [4]. Type II odontoid fractures represent the majority of all odontoid fractures in the elderly population [5,6]. Type II odontoid fractures Epidemiologic data indicate that the incidence of type II fractures follows a bimodal age distribution. Young adults characteristically sustain such fractures from high-impact collisions such as motor vehicle accidents. In contrast, elderly individuals are susceptible, because of increasing frailty and comorbidities, to sustaining their fractures following simple falls [7,8]. Recently, a steady increase in the incidence of geriatric type II fractures has been observed, most likely secondary to an ever aging population [9]. However, the choice between surgical intervention and non-surgical intervention for these fractures in the elderly population remains contentious [10]. Surgical intervention Recent advances in surgical techniques and improvements in surgery safety have resulted in a significantly increased proportion of elderly patients undergoing surgical intervention in the last 2 decades [9]. Briefly, surgical intervention for type II fractures involves either screw or rod fixation via either an anterior or posterior approach. However, surgery in general is more likely to be hazardous in the elderly population because of the presence of significant underlying comorbidities [11]. Additionally, the increased prevalence of osteoporosis observed in this population complicates stability of surgical fixation [12]. Consideration of the postoperative period is also required. Episodes of postoperative delirium are common in the elderly population, with reports of >30% incidence in patients >70 years old undergoing cervical spine surgery [13], particularly if an intensive care unit stay is required [14]. In those patients

aged 80 years or older, the risk of severe postoperative medical complications can be up to 67%, with a mean length of 9 days on intensive care unit and a 6% 30-day mortality [15]. Non-surgical intervention Conversely, non-surgical interventions involve splinting the neck using an external orthosis. Advantages include a less invasive procedure and a consequently shorter hospital stay. However, external immobilization across the neck for time periods up to 12 weeks may be required. This significantly restricts mobility in a population likely to already be limited in mobility and subsequently adversely impacts on the patient’s quality of life. Although concerns regarding immobilization for a prolonged time period in a predominately supine position are well founded (as this predisposes these vulnerable patients to an increased risk of complications such as aspiration pneumonia and cardiopulmonary arrest [16]) these concerns may be overstated because such patients can be mobilized immediately after a hard collar has been fitted. Rationale for this review Best medical practice demands reliable guidelines to inform subsequent patient care. Definitive conclusions regarding the superiority of either surgical or non-surgical intervention in this patient population are lacking. Recently, two metaanalyses addressing this issue have been published [17,18]. Both have tentatively favored surgical over non-surgical intervention but the inclusion of some data for other fracture types, particularly type III, and, in the most recent of these, omission of pertinent papers constitute limitations of these papers that undermine this conclusion. This review differs from previous reviews by comprehensively identifying current published evidence regarding strictly only data on type II fractures, as well as including newly published studies on this topic. A key determinant of the effectiveness of an intervention is the all-cause mortality benefit provided in both the short and long term. Additionally, fracture union can be used as a measure of intervention success. These three outcomes were chosen as pragmatic indicators of intervention effect. Therefore, this systematic review aims to synthesize the current published literature comparing outcomes following surgical and non-surgical interventions for type II odontoid fractures in the elderly population (≥65 years old). The primary outcomes are mortality at short-term follow-up, mortality at long-term follow-up, and radiological union.


1923

Methods

Primary outcomes

Study identification

Mortality at short-term follow-up was defined as death by ≤3 months following intervention. Mortality at long-term follow-up was defined as death by the study endpoint, regardless of a fixed or variable length of individual patient follow-up within a study. Radiological union was defined as reported evidence of osseous union on radiological imaging.

The electronic databases of MEDLINE, MEDLINE InProgress & Other Non-Indexed Citations, Embase, and the Cochrane Central Register of Controlled Trials (CENTRAL) were searched from inception on August 28, 2017. A combination of Medical Subject Headings and text-words for “odontoid process” and “fracture” were used. Full search strategies are outlined in Appendix S1. Results were limited to the English language. No limits on publication date were imposed. Backward- and forward-citation searching of included articles were performed using the reference lists of included studies and Institute for Scientific Information Web of Science, respectively. Eligibility criteria Full-text papers of which the title and abstract met the eligibility criteria (Table 1) were rigorously assessed to determine inclusion. This review was limited to trials that included data comparing surgical intervention with non-surgical intervention in elderly participants, defined pragmatically as ≥65 years old, who were diagnosed with type II (Anderson and D’Alonzo classification) odontoid fractures. An additional limit was the reporting of data for both intervention arms on at least one of our three primary outcomes, which are defined in the following section. Assessment of eligibility was performed in a blinded manner independently by two authors (DPS and AKD) and disagreements were resolved by discussion. Excluded studies and the reasons for exclusion are listed in Appendix S2.

Data extraction A data extraction sheet, based on the Cochrane Consumers and Communication Review Group’s data extraction template, was formulated (Table 2). Data extraction was performed independently by two authors (DPS and AKD). Interventions were classified as either “surgical” or “nonsurgical,” in line with our research question. For studies with only patient-level data, extraction was performed on an intention-to-treat basis. Statistical analysis Mortality rates and radiological union rate, all being dichotomous data, are reported as risk ratios (RRs) with 95% confidence intervals. These were calculated for each study only if outcomes were comparable between intervention arms. An RR less than 1 favored surgical intervention, whereas an RR more than 1 favored non-surgical intervention. Chisquare and I2 statistics, to assess the presence and extent of statistical heterogeneity, respectively, were calculated for each outcome if studies were homogeneous enough to merge. This was based on subjective judgment of methodological heterogeneity (including length of follow-up) and clinical

Table 1 PICOS table detailing the inclusion and exclusion criteria

Participants

Interventions

Comparators Outcomes

Study design

Publications

Inclusion criteria

Exclusion criteria

Both of the following: • Extractable data for all patients aged ≥65 years • Extractable data for all patients with type II odontoid fracture • Both surgical and non-surgical interventions studied

Either of the following: • Data for only patients aged ≥65 years not extractable • Data for only patients with a type II fracture not extractable Either of the following: • Only surgical intervention(s) studied • Only non-surgical intervention(s) studied

• Surgical versus non-surgical interventions Reporting of at least one of the following in both intervention arms: • Mortality at short-term follow-up • Mortality at long-term follow-up • Radiological union rate Either of the following: • Randomized controlled trial • Quasi-experimental study • Observational study • English language

Either of the following: • No report of any primary outcome • Report of a primary outcome in only one intervention arm

Either of the following: • Case report • Secondary research (review or meta-analysis of primary research) Either of the following: • Conference abstract • Letter, comment, or note

This table was formulated using the PICOS (participants, interventions, comparators, outcomes, and study design) approach detailed in the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement [19]. An additional category, “Publications,” was added to encompass the language restrictions.

1924


Table 2 Details of the data extraction form Domain Characteristics of the trial

Eligibility criteria of the trial Type of intervention

Type of outcome measure

Extracted information • • • • • •

Name of first author Year of publication Study design Study setting Number of participants included Number of participants lost to follow-up in each intervention arm • Reported inclusion criteria • Reported exclusion criteria • Specific method of surgical intervention • Number of patients undergoing surgical intervention • Specific method of non-surgical intervention • Number of patients undergoing non-surgical intervention For each intervention arm: 1. Mortality at short-term follow-up • Length of short-term follow-up • Mortality rate at short-term follow-up 2. Mortality at long-term follow-up • Length of long-term follow-up (either predetermined study endpoint or mean, standard deviation, and range) • Mortality rate at long-term follow-up 3. Radiological union rate • Time point at which radiological union assessed • Radiological modality used to assess union • Radiological union rate • Radiological non-union rate • Number of patients deceased before measurement • Number patients lost to follow-up before measurement • Number of patients for which union status was unspecified

heterogeneity. A p value of >.10 in relation to the chisquare statistic signified statistical homogeneity. Stratification of I2 results was based on recommendations by the Cochrane Group [20]. A judgment of the statistical, clinical, and methodological heterogeneity was performed to determine whether study results should be pooled to calculate an aggregate statistic and, if so, whether a fixed- or randomeffects model should be used. Quality assessment Methodological quality The Downs & Black checklist, recommended by the Cochrane Collaboration [20], was used to assess the methodological quality of included non-randomized studies at a study level [21]. Scoring was performed independently by two authors (DPS and AKD), and disagreements were resolved by discussion. The checklist comprises five domains: reporting quality (10 items), external validity (3 items), internal validity with a focus on bias (7 items), internal validity with a focus on confounding (6 items), and statistical power (1 item). A higher score signifies higher quality in that domain.

No validated classification exists to stratify the degree of methodological quality within a domain or of a study as a whole. Therefore, scores were displayed within domains, together with the mean score across studies within that domain, and a narrative synthesis of these scores was performed. Publication bias A funnel plot and subsequent tests for funnel plot asymmetry to identify publication bias were planned for outcomes with data from at least 10 studies [20,22]. Results Study selection A Preferred Reporting Items for Systematic Reviews and Meta-Analyses flowchart detailing the process of study selection is shown in Fig. 1. In total, 12 studies were included in this systematic review. Pertinent details of the included studies are shown in Table 3. Characteristics of included studies Study details All 12 included studies were non-randomized cohort studies; 10 were retrospectively performed [23,24,26–28, 30–34] with the two remaining were performed prospectively [25,27]. All but two of the studies were performed in either United States [23,25,27–31,33,34] or Canada [29,32,33], with the remaining two studies performed in Switzerland [26] and Austria [24]. Two studies were multicentric [29,33] (Table 3). Participants A total of 1,098 participants were involved in the included studies, comprising 483 patients undergoing surgical intervention and 615 patients undergoing non-surgical intervention. Age restrictions for individual studies varied: eight included patients aged ≥65 [24,26–31,34], one included patients aged ≥70 [32], and three included patients aged ≥80 [23,25,33]. The mean age of patients, across the nine studies (n=903) in which this was reported, was 81.9 years. In the surgically managed cohorts (n=322) the mean age was 80.1 years and in the non-surgically managed cohorts (n=425) the mean age was 83.2 years across the studies where this demographic was reported. Only four studies reported a statistical comparison of age between the surgically and non-surgically managed cohorts, two demonstrating a statistically significant difference [25,27], whereas the other two demonstrating no statistically significant difference [24,33]. Of the 12 studies, 8 used comorbidity scores used to assess preinjury comorbidities: 3 studies used the Charlson Comorbidity Index [27,30,31], 2 studies used the American Society of Anesthesiologists physical status classification system [24,26], 1 study used the Cumulative Illness Rating Scale for Geriatrics [32], 1 study used the Short Form-36 version 2 [29], and 1 study used a non-validated score [33]. Of these, five statistically compared scores between the


1738 citations identified through Embase Search

428 citations identified through MEDLINE In-Progress & Other Non-Indexed Citations Search

32 citations identified through CENTRAL Search

Identification

1325 citations identified through MEDLINE Search

1925

2520 citations excluded by title and abstract screen

Titles and abstracts of 2620 discrete citations screened

616 non-discrete citations found via backward- and forwardcitation searching

Screening

84 citations excluded following full-text assessment

Full text of 96 citations assessed

0 additional eligible citations found via hand-searching

Eligibility Included

12 citations included in systematic review

Fig. 1. PRISMA flowchart detailing the process for the selection of papers. Backward-citation searching was performed using the reference lists of included studies and forward-citation searching was performed using ISI Web of Science. Full details of reasons for exclusion for the 84 citations can be found in Appendix S2. Following the independent title and abstract screen of 2,620 citations, there were 254 disagreements (9.7%). Following the independent screen of 616 citations found via backward- and forward-citation searching, there were 6 disagreements (1.0%). Following the independent full-text review of 96 full-text papers, there were 7 disagreements (7.3%). All disagreements were resolved through discussion. CENTRAL, Cochrane Central Register of Controlled Trials; ISI, Institute for Scientific Information; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

surgical and non-surgical cohorts with none demonstrating a significant difference [24,26,29,30,33]. Exclusion criteria varied considerably among studies (Table 3). Eleven studies investigated odontoid fractures of only type II [24–34]; the remaining study included both type II and III fractures but the availability of comprehensive patient-level data allowed isolation of patients with type II fractures [23]. Interventions Surgical interventions included a variety of techniques using either anterior or posterior approaches. Non-surgical interventions included halo-vest immobilization, variations of cervical collar, or solely skeletal traction. In two studies (n=366), the specific surgical or non-surgical intervention was unspecified [27,31] (Table 4). The allocation of treatment was based on surgeon and patients choice in eight studies (n=665) [24–26,29,31–34] and was not reported in two studies (n=359) [27,30]. In the remaining two studies (n=74), this decision was based on the degree of displacement of the odontoid process [23,28].

Mortality at short-term follow-up (≤3 months) Eleven studies (n=939) facilitated extraction of data on mortality at a time point within 3 months; in six of these (n=334), this was at 3-month follow-up [23,26,28–32]. The RRs of these six studies were varied, ranging from 0.18 to 4.17. The 95% confidence intervals of these RRs were wide, with all overlapping the threshold of no effect (Fig. 3). Moderate statistical heterogeneity between studies assessing mortality at 3-month follow-up was detected (chi-square=9.89, degrees of freedom=5, p=.08; I2=49%). Therefore, an aggregate statistic was not calculated. Two studies (n=433) reported 30-day mortality [25,27] (Table 5). One study demonstrated a significantly higher survival in the surgically managed cohort at this time point [27], whereas the other demonstrated no significant difference between groups [25]. One study (n=80) reported 50-day mortality [24] (Table 6) that showed a trend toward reduced mortality in the non-surgically managed cohort. Three studies (n=203) reported “in-hospital mortality” [25,33,34] (Table 7).

Quality assessment

Mortality at long-term follow-up

Methodological quality scores on the Downs & Black checklist for each study are shown in Fig. 2A to E. Full scoring details are shown in Appendix S3. Publication bias was not assessed as less than 10 studies were identified for each outcome.

Six studies (n=822) facilitated extraction of data on mortality at long-term follow-up; three (n=426) reported mortality at a predetermined study endpoint [25,29,31] (Table 8), whereas the remaining three (n=396) reported mortality at

1926

Table 3 Characteristics of included studies Setting

Study design

LoE

Inclusion criteria

Joestl et al. (2016) [24]

Austria

Retrospective cohort

3

Graffeo et al. (2016) [25]

USA

Prospective cohort

3

Scheyerer et al. (2013) [26]

Switzerland


3

Chapman et al. (2013) [27]

USA


3

Molinari et al. (2013) [28]

USA


3

Vaccaro et al. (2013) [29]

USA, Canada

Prospective cohort

3

France et al. (2012) [30]

USA


3

Schoenfeld et al. (2011) [31]

USA


3

Chaudhary et al. (2010) [32]

Canada


3

Smith et al. (2008) [33]

USA, Canada


3

• Age >80 • Type II odontoid fracture • Acute fracture

Kuntz et al. (2000) [34]

USA


3

Hanigan et al. (1993) [23]

USA


3

• • • •

• • • • • • • • • • • • • • • • • •

Age ≥65 ASA score ≥2 Type II odontoid fracture Treated with either anterior screw fixation or halo vest Follow-up of at least 5 years from injury Age ≥80 Type II odontoid fracture Age >65 Type II odontoid fracture Initial XR or CT Age ≥65 Type II odontoid fracture Initially evaluated in emergency room setting then admitted Age ≥65 Type II odontoid fracture Acute fracture Age ≥65 Type II odontoid fracture not longer than 90 days old

• • • • • • • •

Age ≥65 Type II odontoid fracture Age ≥65 Type II odontoid fracture Radiographic evidence of acute fracture documentation Age ≥70 Type II odontoid fracture Treated with either rigid cervical orthosis or surgery

Age ≥65 Type II odontoid fracture Age ≥80 Odontoid fracture (any type)

Exclusion criteria • Type III odontoid fracture • Incomplete dataset (relevant clinical or radiographical data missing) • Penetrating injury • Congenital anomaly of the cervical spine • Halo rigid external fixation • Additional osseous lesions of cervical spine on initial radiography • Presenting delayed to outpatients

None specified

• Previous fracture treatment • Substantial cognitive impairment • Not able to provide information on primary outcome variables (NDI and SF-36v2 scores) • Pathologic fracture None specified

• Treated with halo vest • Displacement >4 mm • Posteriorly displaced fractures • Neurologic compromise • Multilevel cervical spine injury • Associated spine fractures • Medical records not locatable • Documented neurologic injury • Chronic fracture based on radiology None specified • Fracture associated with neoplasm

ASA, American Society of Anesthesiologists; XR, plain X-ray radiograph; CT, computed tomography; NDI, Neck Disability Index; SF-36v2, Short Form-36v2; LoE, level of evidence. Level of evidence was assessed with guidance published by the Oxford Centre for Evidence-based Medicine, updated in 2011 [35]. Absence of reporting of eligibility criteria was denoted as “None specified.”


Author


1927

Table 4 List of specific interventions Author (year)

Intervention arm

N

Joestl et al. (2016) [24] Graffeo et al. (2016) [25]

Surgical Non-surgical Surgical

32 48 17

Scheyerer et al. (2013) [26]

Non-surgical Surgical

94 33

Non-surgical Surgical Non-surgical Surgical Non-surgical Surgical

14 165 157 25 33 101

Non-surgical

58

Surgical

12

Non-surgical

25

Schoenfeld et al. (2011) [31]

Surgical Non-surgical

44 112

Chaudhary et al. (2010) [32]

Surgical

11

Smith et al. (2008) [33]

Non-surgical Surgical

9 32

Non-surgical

40

Kuntz et al. (2000) [34]


6 14

Hanigan et al. (1993) [23]


5 11

Chapman et al. (2013) [27] Molinari et al. (2013) [28] Vaccaro et al. (2013) [29]

France et al. (2012) [30]

Specific intervention techniques Anterior screw fixation (n=32) Halo-vest immobilization (n=48) Posterior C1–C2 segmental polyaxial screw and rod fixation according to Harms (n=15); Anterior fusion using 1-screw (n=2) Hard collar orthosis (n=94) Posterior C1–C2 arthrodesis including a screw-rod fixation system according to Harms (n=16); Direct anterior screw fixation (n=17) Soft collar orthosis (6 weeks) (n=14) Not specified (n=165) Not specified (n=157) Posterior fusion including C1–C2 (n=25) Rigid cervical collar (12 weeks) (n=33) Anterior odontoid screw fixation (n=12); Segmental posterior C1–C2 screw-rod fixation (n=80); Posterior transarticular screw fixation (n=7); Brooks fusion (n=1); Occipitocervical fusion (n=1) Soft collar (n=5); Rigid collar (n=47); Halo-vest immobilization (n=6) Anterior screw fixation (n=7); Posterior C1–C2 fusion (n=5) Cervical collar (n=9); Halo-vest immobilization (n=16) Not specified (n=44) Rigid cervical orthosis (n=84); Halo-vest immobilization (n=28) Odontoid screw fixation (n=not specified); Transarticular screw fixation (n=not specified) Rigid cervical orthosis (3 months) (n=9) Anterior approach (n=10) Posterior approach (n=22) Halo-vest immobilization (n=16) Cervical orthosis (n=24) Posterior C1–C2 transarticular screw fixation with a modified Gallie fusion (n=6) Halo-vest immobilization (n=8); Minerva brace (n=3); Miami J collar (n=3) Posterior cervical wiring of C1–C2 and fusion (n=5) Collar (n=9); Halo-vest immobilization (n=1); Skeletal traction (n=1)

N=sample size. If specific surgical or non-surgical interventions were not detailed within the paper, these were labeled as “Not specified.” The number in parentheses following each intervention represents the number of patients undergoing that specific intervention.

mean follow-up [23,27,28] (Table 9). A tendency toward a reduced mortality at long-term follow-up favoring surgical intervention was observed in three studies (n=637) [27,29,31], whereas in one study (n=58) [28] a slight favor toward nonsurgical intervention was observed. One study demonstrated an equivalent mortality rate between both groups [25]. An RR was not calculated for the remaining study as the mean length of follow-up between the surgical and non-surgical arms differed substantially [23]. Chi-square and I2 statistics were not calculated because of interstudy methodological heterogeneity, particularly in relation to the significant variation of follow-up length.

Radiological union rate Seven studies (n=400) assessed radiological union rates (Table 10). Four studies failed to report the time point of union assessment [23,26,29,34]. Of the remaining three, one reported union assessment at “a minimum of 3 months,” but lacked any further clarification [32], one reported union assessment at patients’ final follow-up appointment, which was of variable length [28], and one reported union assessment at either 3- or 12-month follow-up; although, it was not clear which [24]. Imaging modality used to assess union was plain radiography in the majority. In three studies, the union status

Hanigan et al., 1993


Kuntz et al., 2000 Hanigan et al., 1993

0 Smith et al., 2008

1

Chaudhary et al., 2010

2

Schoenfeld et al., 2011

3

France et al., 2012

4

Vaccaro et al., 2013

7

Molinari et al., 2013

D

Chapman et al., 2013

5 Joestl et al., 2016


Kuntz et al., 2000

Smith et al., 2008



France et al., 2012




Scheyerer et al., 2013

B


Internal Validity (Bias) Graffeo et al., 2016

Quality score (out of 3)

10 9 8 7 6 5 4 3 2 1 0

Graffeo et al., 2016

6



Kuntz et al., 2000

Smith et al., 2008



France et al., 2012





Reporting Quality

Joestl et al., 2016

Kuntz et al., 2000

Kuntz et al., 2000


France et al., 2012




Smith et al., 2008

0

Smith et al., 2008

1 Chaudhary et al., 2010

Statistical Power



France et al., 2012




E Scheyerer et al., 2013

Joestl et al., 2016 Graffeo et al., 2016


A


Joestl et al., 2016 Graffeo et al., 2016


C

Graffeo et al., 2016

Joestl et al., 2016


1928 D.P. Sarode and A.K. Demetriades / The Spine Journal 18 (2018) 1921–1933

External Validity

3

2

1

0

Internal Validity (Confounding)

6

5

4

3

2

1

0

Fig. 2. A to E: Methodological quality scores on the Downs & Black checklist separated by assessment domains. Separation of checklist items into domains was performed as suggested by the original publication of the checklist. Full details of the scoring are displayed in Appendix S3. The red dotted lines represent the mean score across the 12 studies within each domain.


1929

Fig. 3. Mortality at 3-month follow-up. Events represent all-cause mortality within the intervention arm. The aggregate statistic was suppressed due to the reasons explained in text. This figure was created using the RevMan 5.1 software (The Cochrane Collaboration, The Nordic Cochrane Centre, Copenhagen) and modified using Microsoft Paint (Microsoft, Redmond, WA, USA). Stratification of I2 values was as follows: 0% to 40%=no significant heterogeneity; 30% to 60%=moderate statistical heterogeneity; 50% to 90%=substantial statistical heterogeneity; 75% to 100%=considerable statistical heterogeneity. CI, confidence interval.

of some patients was unspecified, ranging from 1 patient to 142 patients [26,29,34]. For all studies, the lack of reporting a fixed time point for assessment made it unclear whether union was assessed at the same time point in both

intervention arms. This precluded calculation of RRs and statistical heterogeneity statistics as they would be misleading to display.

Discussion

Table 5 Mortality at 30-day follow-up

Discussion of findings

Author (year)

Intervention arm

N

Mortality at 30-day follow-up

Graffeo et al. (2016) [25] Chapman et al. (2013) [27]

Surgical Non-surgical Surgical Non-surgical

17 94 165 157

4 (23.5%) 25 (26.6%) 11 (6.7%) 35 (22.3%)

Risk ratio [95% CI] 0.85 [0.25–2.85] 0.30 [0.16–0.57]

N, sample size; CI, confidence interval. The risk ratio value represents mortality following surgical intervention compared with non-surgical intervention.

Table 6 Mortality at 50-day follow-up

Author (year)

Intervention arm

N

Mortality at 50-day follow-up

Joestl et al. (2016) [24]


32 48

4 (12.5%) 3 (6.3%)

Risk ratio [95% CI] 4.33 [0.91–20.53]


As a whole, the current evidence regarding the best choice of management for geriatric type II fractures is of insufficient quality and homogeneity to produce definitive conclusions. The available literature suggests that although surgical intervention does not appear to confer a mortality benefit at 3 months, reduced mortality rates may be observed at longer intervals. Indeed, a time-dependent effect may explain why literature on this topic, in which no standardized follow-up lengths are defined, is so divided. However, the plethora of study limitations, which will be discussed later, greatly undermines the reliability of this conclusion. Analysis of the radiological union rate was more difficult. Consideration of the time point of radiological union assessment is vital when comparing studies as the outcome measure may represent different stages of healing. However, insufficient time point reporting made it unclear whether radiological union was assessed at the same time point for each intervention arm, introducing both intra- and interstudy heterogeneity. This, therefore, precluded adequate analysis of the radiological union rate.

Table 7 “In-hospital” mortality Author (year)

Intervention arm

N

Mean length of follow-up (days)

Range of follow-up (days)

Mortality at short-term follow-up (%)

Graffeoet al. (2016) [25]

Surgical Non-surgical Surgical Non-surgical Surgical Non-surgical

17 94 32 40 6 14

Unknown Unknown 22.8±28.3 11.2±8.5 14.2±5.3 11.7±6.6

Unknown Unknown Unknown Unknown 5–19 3–24

3 (17.6%) 11 (11.7%) 4 (12.5%) 6 (15%) 1 (16.7%) 3 (21.4%)

Smith et al. (2008) [33] Kuntz et al. (2000) [34]

Risk ratio [95% CI] 0.30 [0.08–1.12] 0.78 [0.10–6.05] 0.83 [0.26–2.70]

N, sample size; CI, confidence interval. Values in the mean length of follow-up column represent mean±standard deviation. The risk ratio value represents mortality following surgical intervention compared with non-surgical intervention. Data that were not extractable was labeled as “Unknown.”

1930


Table 8 Mortality at long-term follow-up (fixed length of follow-up) Author (year)

Intervention arm

N

Long-term follow-up (year)

Mortality at long-term follow-up (%)

Graffeo et al. (2016) [25]


17 94 101 58 44 112

1 1 1 1 3 3

7 (41.2%) 39 (41.5%) 14 (14%) 15 (26%) 11 (25%) 50 (45%)

Vaccaro et al. (2013) [29] Schoenfeld et al. (2011) [31]

Risk ratio [95% CI] 0.99 [0.35–2.82] 0.54 [0.28–1.03] 0.56 [0.32–0.97]


The lack of appropriate evidence to allow synthesis of outcome measures on this topic has been reported previously [36–38]. Two meta-analyses have recently attempted to address the present research question [17,18]. Schroeder et al. demonstrated a reduced short- and long-term mortality following surgical intervention but stated the majority of included studies were at a high risk of bias [17]. Their inclusion of single-arm studies, which is also a feature of a previous meta-analysis [36], is also likely to markedly in-

crease the clinical and methodological heterogeneity. This may explain the discrepancy in the short-term mortality conclusions compared with our review. Furthermore, there was a contribution from type III fractures in two of their included studies [39,40]. Similar to this review, concern was expressed regarding the vast heterogeneity of union status reporting [17]. Additionally, Yang et al. also demonstrated similar levels of statistical heterogeneity in terms of mortality in their select

Table 9 Mortality at long-term follow-up (variable length of follow-up) Author (year)

Intervention arm

N

Mean length of follow-up (days)

Range of follow-up (days)

Mortality at long-term follow-up (%)

Chapmanet al. (2013) [27]


165 157 25 33 5 11

851 648 414 447 1,344 342

0–2,456 1–2,565 8–1,440 60–1,440 72–2,160 4–1,530

62 (38%) 80 (51%) 5 (20%) 4 (12%) 1 (20%) 8 (72.7%)

Molinari et al. (2013) [28] Haniganet al. (1993) [23]

Risk ratio [95% CI] 0.74 [0.57–0.95] 1.65 [0.49–5.52] Not calculated

N, sample size; CI, confidence interval. Follow-up lengths reported in months were converted to days (30 days per month) to facilitate simpler display of data. Data on mean length and range of follow-up were rounded to integers. The risk ratio value represents mortality following surgical intervention compared with non-surgical intervention.

Table 10 Radiological union rate Author (year)

Intervention arm

N

Joestl et al. (2016) [24] Scheyerer et al. (2013) [26] Molinari et al. (2013) [27] Vaccaro et al. (2013) [29] Chaudhary et al. (2010) [32] Kuntz et al. (2000) [34] Hanigan et al. (1993) [23]

Surgical Non-surgical Surgical Non-surgical Surgical Non-surgical Surgical Non-surgical Surgical Non-surgical Surgical Non-surgical Surgical Non-surgical

32 48 33 14 25 33 101 58 11 9 6 14 5 11

Time point union assessed

Imaging modality

Radiological union (%)

Radiological non-union (%)

Deceased at measurement (%)

Lost to follow-up (%)

Unspecified (%)

Either 3 months or 12 months Unknown

XR+CT

26 (81.3%) 34 (70.8%) 17 (51.5%) 0 7 (28.0%) 2 (6.1%) Unknown Unknown 7 (63.6%) 6 (66.7%) 4 (66.7%) 3 (21.4%) 2 (40.0%) 3 (27.3%)

3 (9.4%) 10 (20.8%) 7 (41.2%) 8 (57.1%) 13 (52.0%) 30 (90.9%) 5 (5.0%) 12 (20.7%) 1 (9.1%) 2 (2.2%) 0 1 (7.1%) 3 (60.0%) 3 (27.3%)

3 (9.4%) 4 (8.3%) 4 (23.5%) 0 4 (16.0%) 0 Unknown Unknown 3 (27.3%) 1 (1.1%) 1 (16.7%) 3 (21.4%) 0 5 (45.5%)

0 0 4 (23.5%) 6 (42.9%) 1 (4.0%) 1 (3.0%) Unknown Unknown 0 0 1 (16.7%) 1 (7.1%) 0 0

0 0 1 (3.0%) 0 0 0 96 (95.0%) 46 (79.3%) 0 0 0 6 (42.9%) 0 0

XR+CT

Ultimate follow-up Unknown

XR Unknown

≥3 months

XR

Unknown

XR

Unknown

XR or CT

N, sample size; XR, plain X-ray radiograph; CT, computer tomography. Values represent absolute numbers and percentage of deaths within the intervention arm. Ultimate follow-up was defined as the final follow-up appointment after which the patient was lost to follow-up. Data that were not extractable was labeled as “Unknown.”


group of included studies and no difference in mortality between interventions [18]. Interestingly, they found no statistical heterogeneity regarding the non-union outcome [18]. However, comparability between this present review and that of Yang et al. is undermined by the fact that the latter had a significant contribution from type III fractures [41–43] and omitted some papers included here [26–30]. Limitations of included studies A significant hindrance to assessing the reliability of evidence in the included studies was the variable reporting quality of pertinent study details (Fig. 2A). These details included eligibility criteria, important confounding factors (other coexisting spinal fractures, for example), and incomplete outcome data for all patients. Use of a reporting checklist, such as that published by the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) initiative [44], will enable a transparent research process and will aid in future synthesis of evidence. Reporting aside, the most significant limitation of these studies is that, in the absence of randomization, the baseline characteristics of patients between intervention arms are unlikely to be comparable as treatment decisions were mostly based on surgeon and patient choice. Indeed, “internal validity (confounding)” scores were universally low indicating poor baseline comparability between intervention arms in all studies (Fig. 2D). Surgical measures are not undertaken lightly in the elderly population, and therefore, surgical cohorts may be younger and fitter. The studies could be interpreted in the context of “confounding by indication” in which intervention stratification within studies represents real-life intervention stratification. The subjectivity of such a decision, however, is likely to introduce interinstitution variability, limiting the external validity of this approach. Poor generalizability of results from the included papers is suggested by the external validity scores, with 7 of the 12 studies scoring ≤1 of 3 (Fig. 2B). This hinders the synthesis of generalized recommendations to ultimately inform clinical practice in a meaningful manner. Statistical adjustments for confounding factors in the data analysis provide a method of controlling differences between intervention arms in non-randomized studies. However, this method was used in only three studies, with the remaining studies not controlling for confounding factors. Although the use of such statistical models can be challenging, implementation of these would bolster the reliability of non-randomized evidence if done so correctly. This is particularly important in the context of a clinical issue in which randomized controlled trials (RCTs) are difficult to perform and are severely lacking. As demonstrated by the “internal validity (bias)” scoring on the Downs & Black checklist, all studies were at a high risk of overall bias (Fig. 2C). Although blinding is generally difficult to perform in relation to surgical studies, this was lacking in all 12 studies and therefore may introduce bias sec-

1931

ondary to selective reporting. A protocol was not published before 10 of the studies, further increasing this risk. Most studies involved small sample sizes, with only one offering an accompanying power calculation (Fig. 2E). Although sample size calculations for non-randomized studies are more difficult to perform, published methods exist [45]. Thus, whether these studies are sufficiently powered to reliably detect a significant effect is unknown as small sample sizes may mask statistically significant effects. Additionally, some studies used an “as-treated” analysis, others an “intention-to-treat” analysis, and the remainder did not specify the type of analysis used. An “as-treated” analysis together with absence of reporting of number of patients who were crossed over severely impedes reliable analysis. Limitations of the review process Considering that the management of geriatric type II odontoid fractures constitutes a growing area of controversy, surprisingly, few studies addressing this issue were identified, none of which were RCTs. However, limitations of the present search strategy may prevent identification of all the available evidence. First, although conference abstracts are a rich source of research, these were excluded as determining the validity of reported results is challenging. Second, limiting the search to only English articles undoubtedly restricts the scope of the search. Finally, unpublished evidence was not sought in this review. Classification of treatments as either “surgical” or “nonsurgical” intervention, as performed here, overlooks any differences between specific techniques within these categories. For example, comparing anterior with posterior surgical approaches, the former is less invasive than the latter but it has been suggested that the former results in lower union rates [46]. A larger dataset is required to allow subgroup analysis to adequately critique individual interventions. It was decided here to perform only a narrative synthesis on the grounds of the significant clinical, methodological, and statistical heterogeneity demonstrated. This differs from previous meta-analyses, which synthesized an aggregate statistic. However, it was believed that with the data identified by this present review, the display of an aggregate statistic would be misleading and would distract from the fact that more highquality evidence is required before drawing reliable, statistically sound conclusions. Conclusions and future directions In summary, the main finding of our review is that interstudy heterogeneity and poor quality of the existing evidence preclude reliable conclusions being drawn. This constitutes a significant barrier in translating findings from individual studies to robust clinical guidelines. In future observational studies, reporting using the framework suggested by the STROBE initiative will prove invaluable for reliable evidence synthesis [39].

1932


Although RCTs represent the gold standard, only a single 2007 report [47] of such a study has been identified, reflecting the difficulty in performing such a trial. This interim report, however, required 20 additional patients to reach their target sample size but there has been no further publication. Contact with the authors has revealed there has been no further progress with the clinical trial since this report but dissemination of the incomplete trial results would be considered in the near future. Nevertheless, there is a distinct lack of RCTs addressing this important issue, the presence of which would greatly bolster the current evidence base. An initial assessment of frailty and medical comorbidities with suitable patients subsequently randomized to surgical or non-surgical treatment would be a practical, and clinically applicable, approach and would help in overcoming some of the issues faced in performing such an RCT. In terms of ongoing research, the Uppsala Study on Odontoid Fracture Treatment in the elderly (USOFT), which is aiming to compare posterior fusion with rigid cervical collar in 50 patients aged 75 years or older suffering from acute type II odontoid fractures, is currently underway. The protocol for this RCT has been registered in ClinicalTrials.gov (ID: NCT 02789774) and the trial is expected to reach completion in February of 2019. It will be interesting to see whether the results of future trials will facilitate meaningful synthesis in the future. Nevertheless, current evidence is not sufficient to resolve this ongoing debate and in the era of evidence-based medicine, there is a need for further higher quality evidence to guide the management of this patient population. Supplementary material

[10]

[11]

[12]

[13]

[14] [15]

[16]

[17]

[18]

[19]

[20]

[21]

Supplementary material related to this article can be found at https://doi.org/10.1016/j.spinee.2018.05.017. [22]

References [1] Ryan MD, Henderson JJ. The epidemiology of fractures and fracturedislocations of the cervical spine. Injury 1992;23:38–40. [2] Venkatesan M, Northover JR, Wild JB, Johnson N, Lee K, Uzoigwe CE, et al. Survival analysis of elderly patients with a fracture of the odontoid peg. Bone Joint J 2014;96-B:88–93. [3] Baaj AA, Uribe JS, Nichols TA, Theodore N, Crawford NR, Sonntag VK, et al. Health care burden of cervical spine fractures in the United States: analysis of a nationwide database over a 10-year period. J Neurosurg Spine 2010;13:61–6. [4] D’Alonzo AL. Fractures of the odontoid process of the axis. J Bone Joint Surg 1974;56A:1663–74. [5] Ryan MD, Taylor TKF. Odontoid fractures in the elderly. J Spinal Disord 1993;6:397–401. [6] Malik SA, Murphy M, Connolly P, O’Byrne J. Evaluation of morbidity, mortality and outcome following cervical spine injuries in elderly patients. Eur Spine J 2008;17:585–91. [7] Pepin JW, Bourne RB, Hawkins RJ. Odontoid fractures, with special reference to the elderly patient. Clin Orthop Relat Res 1985;193:178–83. [8] Clark CR, White AA 3rd. Fractures of the dens. A multicenter study. J Bone Joint Surg Am 1985;67:1340–8. [9] Smith HE, Kerr SM, Fehlings MG, Chapman J, Maltenfort M, Zavlasky J, et al. Trends in epidemiology and management of type II odontoid

[23] [24]

[25]

[26]

[27]

[28]

fractures: 20-year experience at a model system spine injury tertiary referral center. J Spinal Disord Tech 2010;23:501–5. Ryken TC, Hadley MN, Aarabi B, Dhall SS, Gelb DE, Hurlbert RJ, et al. Management of isolated fractures of the axis in adults. Neurosurgery 2013;72(Suppl. 2):132–50. Harris MB, Reichmann WM, Bono CM, Bouchard K, Corbett KL, Warholic N, et al. Mortality in elderly patients after cervical spine fractures. J Bone Joint Surg Am 2010;92:567–74. Clarke BL, Ebeling PR, Jones JD, Wahner HW, O’Fallon WM, Riggs BL, et al. Changes in quantitative bone histomorphometry in aging healthy men. J Clin Endocrinol Metab 1996;81:2264–70. Ushida T, Yokoyama T, Kishida Y, Hosokawa M, Taniguchi S, Inoue S, et al. Incidence and risk factors of postoperative delirium in cervical spine surgery. Spine 2009;34:2500–4. Robinson TN, Eiseman B. Postoperative delirium in the elderly: diagnosis and management. Clin Interv Aging 2008;3:351–5. Ryang YM, Török E, Janssen I, Reinke A, Buchmann N, Gempt J, et al. Early morbidity and mortality in 50 very elderly patients after posterior atlantoaxial fusion for traumatic odontoid fractures. World Neurosurg 2016;87:381–91. Majercik S, Tashjian RZ, Biffl WL, Harrington DT, Cioffi WG. Halo vest immobilization in the elderly: a death sentence? J Trauma 2005;59:350–6. Schroeder GD, Kepler CK, Kurd MF, Paul JT, Rubenstein RN, Harrop JS, et al. A systematic review of the treatment of geriatric type II odontoid fractures. Neurosurgery 2015;77(Suppl. 4):S6–14. Yang Z, Yuan ZZ, Ma JX, Ma XL. Conservative versus surgical treatment for type II odontoid fractures in the elderly: grading the evidence through a meta-analysis. Orthop Traumatol Surg Res 2015;101:839–44. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gøtzsche PC, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. BMJ 2009;339:b2700. Higgins JPT, Green S. Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [updated March 2011]. The Cochrane Collaboration. 2011. Available at: www.cochrane-handbook.org. Accessed December 15, 2015. Downs SH, Black N. The feasibility of creating a checklist for the assessment of the methodological quality both of randomised and non-randomised studies of health care interventions. J Epidemiol Community Health 1998;52:377–84. Sterne JA, Gavaghan D, Egger M. Publication and related bias in meta-analysis: power of statistical tests and prevalence in the literature. J Clin Epidemiol 2000;53:1119–29. Hanigan WC, Powell FC, Elwood PW, Henderson JP. Odontoid fractures in elderly patients. J Neurosurg 1993;78:32–5. Joestl J, Lang N, Bukaty A, Platzer P. A comparison of anterior screw fixation and halo immobilisation of type II odontoid fractures in elderly patients at increased risk from anaesthesia. Bone Joint J 2016;98B:1222–6. Graffeo CS, Perry A, Puffer RC, Carlstrom LP, Chang W, Mallory GW, et al. Deadly falls: operative versus nonoperative management of type II odontoid process fracture in octogenarians. J Neurosurg Spine 2016;26:4–9. Scheyerer MJ, Zimmermann SM, Simmen HP, Wanner GA, Werner CM. Treatment modality in type II odontoid fractures defines the outcome in elderly patients. BMC Surg 2013;13:54. Chapman J, Smith JS, Kopjar B, Vaccaro AR, Arnold P, Shaffrey CI, et al. The AOSpine North America geriatric odontoid fracture mortality study: a retrospective review of mortality outcomes for operative versus nonoperative treatment of 322 patients with long-term follow-up. Spine 2013;38:1098–104. Molinari WJ 3rd, Molinari RW, Khera OA, Gruhn WL. 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;3:21–32.

D.P. Sarode and A.K. Demetriades / The Spine Journal 18 (2018) 1921–1933 [29] Vaccaro AR, Kepler CK, Kopjar B, Chapman J, Shaffrey C, Arnold P, et al. Functional and quality-of-life outcomes in geriatric patients with type-II dens fracture. J Bone Joint Surg Am 2013;95:729–35. [30] France JC, Powell EN 2nd, Emery SE, Jones DL. Early morbidity and mortality associated with elderly odontoid fractures. Orthopedics 2012;35:e889–94. [31] Schoenfeld AJ, Bono CM, Reichmann WM, Warholic N, Wood KB, Losina E, et al. Type II odontoid fractures of the cervical spine: do treatment type and medical comorbidities affect mortality in elderly patients? Spine 2011;36:879–85. [32] Chaudhary A, Drew B, Orr RD, Farrokhyar F. Management of type II odontoid fractures in the geriatric population: outcome of treatment in a rigid cervical orthosis. J Spinal Disord Tech 2010;23:317–20. [33] Smith HE, Kerr SM, Maltenfort M, Chaudhry S, Norton R, Albert TJ, et al. Early complications of surgical versus conservative treatment of isolated type II odontoid fractures in octogenarians: a retrospective cohort study. J Spinal Disord Tech 2008;21:535–9. [34] Kuntz C 4th, Mirza SK, Jarell AD, Chapman JR, Shaffrey CI, Newell DW. Type II odontoid fractures in the elderly: early failure of nonsurgical treatment. Neurosurg Focus 2000;8:e7. [35] OCEBM Levels of Evidence Working Group. The Oxford Level of Evidence 2. Oxford Centre for Evidence-Based Medicine. Available at: http://www.cebm.net/index.aspx?o=5653. Accessed September 30, 2017. [36] Robinson Y, Robinson AL, Olerud C. Systematic review on surgical and nonsurgical treatment of type II odontoid fractures in the elderly. Biomed Res Int 2014;2014:231948. [37] Pal D, Sell P, Grevitt M. Type II odontoid fractures in the elderly: an evidence-based narrative review of management. Eur Spine J 2011;20:195–204. [38] Elgafy H, Dvorak MF, Vaccaro AR, Ebraheim N. Treatment of displaced type II odontoid fractures in elderly patients. Am J Orthop (Belle Mead NJ) 2009;38:410–16.

1933

[39] Woods BI, Hohl JB, Braly B, Donaldson W 3rd, Kang J, Lee JY. Mortality in elderly patients following operative and nonoperative management of odontoid fractures. J Spinal Disord Tech 2014;27:321–6. [40] Ardeshiri A, Asgarib S, Lemonasa E, Oezkana N, Schlamannc M, Surea U, et al. Elderly patients are at increased risk for mortality undergoing surgical repair of dens fractures. Clin Neurol Neurosurg 2013;115: 2056–61. [41] Müller EJ, Wick M, Russe O, Muhr G. Management of odontoid fractures in the elderly. Eur Spine J 1999;8:360–5. [42] Andersson S, Rodrigues M, Olerud C. Odontoid fractures: high complication rate associated with anterior screw fixation in the elderly. Eur Spine J 2000;9:56–9. [43] Fagin AM, Cipolle MD, Barraco RD, Eid S, Reed JF 3rd, Li PM, et al. Odontoid fractures in the elderly: should we operate? J Trauma 2010;68:583–6. [44] von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP, et al. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ 2007;335:806–8. [45] Johnson ES, Brookhart MA, Myers JA. Study Size Planning, Chapter 9. In: Velentgas P, Dreyer NA, Nourjah P, et al., editors. Developing a Protocol for Observational Comparative Effectiveness Research: A User’s Guide. Rockville (MD: Agency for Healthcare Research and Quality (US); 2013. [Chapter 9]. Available at: http://www.ncbi.nlm.nih.gov/books/NBK126183/. Accessed October 1, 2016. [46] Shen Y, Miao J, Li C, Fang L, Cao S, Zhang M, et al. A meta-analysis of the fusion rate from surgical treatment for odontoid factures: anterior odontoid screw versus posterior C1-C2 arthrodesis. Eur Spine J 2015;24:1649–57. [47] Hurlbert RJ, Fox R, Casha S, Broad R, Jiang H DuPlessis S. Interim analysis on a prospective randomized odontoid fracture clinical trial. Can J Surg 2007;50(3 Suppl.):S10–11.