Peer-Review Reports

Complications Associated with Surgical Treatment of Traumatic Spinal Fractures:

A Review of the Scoliosis Research Society Morbidity and Mortality Database

Brian J. Williams1, Justin S. Smith1, Dwight Saulle1, Christopher P. Ames2, Lawrence G. Lenke3, Paul A. Broadstone4,Alexander R. Vaccaro5, David W. Polly Jr 6, Christopher I. Shaffrey1

-OBJECTIVE: Traumatic spinal fracture is a common indication for surgery,with an associated high incidence of perioperative complications. The literatureprovides a wide range in the incidence of complications. We seek to assess theperioperative morbidity and mortality of surgery for traumatic spinal fracturesand to identify predictors of their occurrence.

-METHODS: We performed a retrospective analysis of all traumatic spinalfracture cases submitted by members of the Scoliosis Research Society from2004 to 2007.

-RESULTS: A total of 108,478 cases were submitted from 2004 through 2007,with 6,706 (6.2%) performed for treatment of traumatic fracture. Twenty-twopercent of patients had preoperative neurological deficits. Intraoperative neu-romonitoring was used in 58% of cases. The overall incidence of complicationswas 6.9%. The perioperative mortality was 0.5%. There were 59 (0.9%) newpostoperative neurological deficits. Multivariate analysis demonstrated preop-erative neurological deficit (P [ .001; odds ratio [OR] 1.449, 95% confidenceinterval [CI] [1.156 to 1.817]) and fusion (P [.001; OR 1.12, 95% CI [1.072 to 1.168])as predictors of complications and use of intraoperative neuromonitoring(P [ .016; OR 1.949, 95% CI [1.13 to 3.361]), and preoperative neurological deficit(P < .001; OR 2.964, 95% CI [1.667 to 5.271]) as predictors of new postoperativeneurological deficits (P < .001).

-CONCLUSIONS: Overall, surgery for the treatment of spinal fractures wasperformed with relatively low incidences of perioperative complications (6.9%)and mortality (0.5%). These data may prove useful for patient counseling andongoing efforts to improve the safety of operative care for patients with spinalfracture.

Key words- Complication- Fracture- Mortality- Spine surgery- Spine trauma

Abbreviations and AcronymsMM: Morbidity and mortalitySRS: Scoliosis Research Society

From the 1University of Virginia Medical Center,Department of Neurosurgery, Charlottesville,

Virginia, USA; 2University of CaliforniaeSan Francisco,Department of Neurosurgery, San Francisco, California, USA;3Department of Orthopaedic Surgery, Washington UniversitySchool of Medicine, St Louis, Missouri, USA; 4Spine SurgeryAssociates, Chattanooga, Tennessee, USA; 5Department ofOrthopaedic Surgery, Thomas Jefferson University,Philadelphia, Pennsylvania, USA; 6Departments ofOrthopedic Surgery and Neurosurgery, University ofMinnesota, Minneapolis, Minnesota, USA

To whom correspondence should be addressed:Justin S. Smith, M.D., Ph.D.[E-mail: jss7f@virgina.edu]

Citation: World Neurosurg. (2014).http://dx.doi.org/10.1016/j.wneu.2013.02.013

Journal homepage: www.WORLDNEUROSURGERY.org

Available online: www.sciencedirect.com

1878-8750/$ - see front matter ª 2014 Elsevier Inc.

INTRODUCTION

Traumatic fracture is a common indicationfor spinal surgery. The surgical treatmentof these injuries remains a difficult clinicalproblem with a high incidence of peri-operative complications (4, 15-17). Thispatient population frequently has associ-ated injuries to other major organ systemsor pre-existing comorbid conditions(e.g. ankylosing spondylitis), whichcontributes to significant perioperativemorbidity (2, 3, 6, 7, 9, 10, 18, 19). Specificfracture patterns and regions are alsoassociated with injuries to other organsystems. For example, cervical spinaltrauma is frequently associated with headinjury and facial fractures (11, 12), whereasthoracolumbar flexion distraction injuriesare associated with abdominal organinjury (1).

All rights reserved.

WORLD NEUROSURGERY- (-): ---, MON

Traumatic spine fractures typicallyoccur due to high energy mechanisms, forexample falls from height or motor vehicleaccidents (5, 6, 16, 17). The patient pop-ulation is typically young (<40 years) witha male predominance (5, 6, 16, 17).Patients often present without a neurolog-ical deficit despite potentially significantfracture to the spinal column (6). Thereare many patterns of spinal fracture,however the most frequently encounteredare compression type injuries (6, 15).Overall injuries vary from minor end platecompression fractures to severe fracturedislocation injuries. Fractures of thecervical spine more commonly presentwith a neurological deficit compared to

TH 2014

thoracolumbar injuries (6). The mostlikely levels of the spine involved arejunctional, including the cervicothoracicjunction, thoracolumbar junction, andatlantoaxial regions (5, 6, 15).The Scoliosis Research Society (SRS)

has been collecting morbidity andmortality (MM) data from its members forover 40 years. The membership of the SRSis predominantly fellowship-trained spinalsurgeons and pediatric orthopedists, andthe cases submitted represent a broadrange of case complexities. The SRS MMdatabase has previously been validatedusing historical controls and comparingcomplication rates of common procedures(20). In the present study, we sought to

www.WORLDNEUROSURGERY.org 1

Table 1. Patient and TreatmentCharacteristics for 6706 Patients withSurgically Treated Traumatic SpinalFractures

PEER-REVIEW REPORTS

BRIAN J. WILLIAMS ET AL. COMPLICATIONS OF TRAUMATIC SPINAL FRACTURE SURGERY

assess the complication rates associatedwith the treatment of fractures and toevaluate for factors associated with theoccurrence of complications in a modernseries of patients.

Characteristic N (%)

Age

Mean � SD 57 � 25

Median, range 61, 1e97

Preoperative neurological deficit 1452 (22)

Revision 339 (5.1)

Fracture level

Occipitocervical 399 (5.9)

Subaxial cervical 910 (13.6)

Thoracolumbar 5283 (78.8)

Not recorded 114 (1.7)

Intraoperative neuromonitoring

Yes 3945 (58)

No 2746 (41)

Not recorded 15 (1)

Minimally invasive 1668 (25)

Kyphoplasty 986 (59)

Vertebroplasty 146 (9)

Mini open 199 (12)

Thoracoscopy 15 (1)

Other 322 (19)

Fusion 3617 (54)

Fusion approach

Posterior 2489 (69)

Intralaminar 1132 (45)

Posterolateral 1253 (50)

Transforaminal lumbarinterbody fusion

53 (2)

Posterior lumbar interbodyfusion

51 (2)

Anterior 659 (18)

Combined 427 (12)

Anterior/posterior 412 (96)

Posterior/anterior/posterior 15 (4)

Not recorded 44 (1)

METHODS

The SRSMMdatabase is the compilation ofsubmissions of SRS members, predomi-nantly within North America. For the yearsthat the data were collected and reviewedfor this report, candidate members wererequired to report their operative cases,including MM data. Full active memberswere encouraged to report their casesas well. Whether candidate membersultimately achieved full membershipwas not impacted by the numbers ortypes of complications reported. Thedatabase is designed to focus on perioper-ative complications rather than long-termor distant complications. Perioperativecomplications typically occurred aroundthe time of surgery.We evaluated all surgical cases from 2004

through 2007 from the SRS MM database.All data had previously been deidentifiedwith regard to patient, surgeon, and insti-tution. The database on which this projectwas based was submitted to an institutionalreview board and was deemed to be exemptfrom institutional review board approvalbased on the use of deidentified data.We identified all cases of traumatic

spinal fracture. We then evaluated forcomplications and associated characteris-tics, including preoperative neurologicalinjury, surgical approach, patient age, andrevision procedure. Specific complicationswere collected in the database with a focuson the intraoperative and immediatepostoperative periods. Examples of theseincluded death, new neurological deficit,wound infection (superficial or deep),pulmonary embolus, deep venous throm-bosis, other pulmonary complications,implant related complications, peripheralnerve deficit, visual deficit, and epiduralhematoma. A category for “other”complications, with free text entry fordescription of the complication, was alsoincluded.Statistical analyses of these data were

performed using SPSS for Windowsversion 19.0 (SPSS Inc., Chicago, Illinois,USA). Continuous variables are presentedas mean with standard deviation and

2 www.SCIENCEDIRECT.com

median with range. Categorical variablesare presented as frequency and per-centage. The c2 or Fisher exact test wasused for categorical variables as appro-priate. The 2-sample unpooled t test withunequal variance was used for analysis of2 means when normally distributed. TheWilcoxon signed rank test was used ifvariables were not normally distributed.All tests were 2-tailed. Univariate analysiswas performed to identify predictors ofcomplications and new neurologicaldeficits, and those with a value ofP < .1 or deemed clinically importantwere included in subsequent multivariatebinomial logistic regression analysis.Factors evaluated with univariate analysisfor complications were: age, preoperativeneurological deficit, revision procedure,fusion, minimally invasive technique,approach, and fracture level. Factorsevaluated with univariate analysis fornew neurological deficit were: age,preoperative neurological deficit, intra-operative neuromonitoring, revision,fusion, approach, and fracture level. Thethreshold for statistical significance wasset at P < .05.

RESULTS

Patient and Treatment CharacteristicsA total of 108,478 cases were submittedfrom 2004 through 2007, with 6706 (6.2%)performed for treatment of traumaticfracture. A summary of the baselinecharacteristics of the patient populationare listed in Table 1. The cases weresubmitted by either active members(encouraged to submit) or candidatemembers (required to submit). Thesubmissions were comprised of 72.2%(4842) active members, candidate 27.4%(1839), and not recorded 0.4% (25).Twenty-two percent of patients hadpreoperative neurological deficits. Thesedeficits were comprised of 143 cases ofcauda equina, 428 complete cord, 548incomplete cord, and 279 nerve rootinjuries. Intraoperative neuromonitoringwas used in 58% of cases. Minimallyinvasive techniques were used in 25% ofcases. Thoracolumbar injuries were themost common (78.8%). A fusion wasperformed in 54% of cases, with a poste-rior approach being the most commonmethod (69%). Among the cases treated

WORLD NEUROSURGERY, http://

without fusion (n ¼ 3089) were 986(31.9%) kyphoplasties, 146 (4.7%) verte-broplasties, 1862 (60.3%) cases with use ofinstrumentation without arthrodesis, and95 (3%) not recorded.

dx.doi.org/10.1016/j.wneu.2013.02.013

Table2.

Complications

for6706

PatientsTrea

tedwith

Trau

matic

Spinal

Frac

ture

Stratifiedby

Level

Fracture

Level

(N)

Age

(years)

Mea

n�

SD

Preope

rativ

eNeu

rologica

lDeficitN(%

)

Complications,N

(%)

Overall

Mortality

Infection

Pulmon

aryDurotom

y

Spinal

Cord

Injury

Epidural

hematom

aMalpositio

ned

implan

tsVe

nous

thrombo

embolis

m

Posterior

Isch

emic

Optic

Neu

ropathy

Other

New

Neu

rologica

lDeficit

SuperficialDee

p

Occipitocervical

(399)

56�

2749

(12.3)

42(10)

6(1.5)

2(0.5)

6(1.5)

1(0.3)

4(1)

4(1)

0(0)

7(1.8)

1(0.3)

0(0)

10*(2.5)

7(1.8)

Subaxial

cervical(910)

44�

31480(53)

72(7.9)

8(0.9)

3(0.3)

12(1.3)

8(0.9)

9(1)

5(0.5)

1(0.1)

16(1.8)

2(0.2)

1(0.1)

2y(0.2)

13(1.4)

Thoracolum

bar

(5283)

59�

22923(17)

352(6.7)

21(0.4)

36(0.7)

81(1.5)

22(0.4)

65(1.2)

12(0.2)

6(0.1)

51(1)

8(0.2)

1(0.01)

31z

39(0.7)

Overall(6706)

57�

251452

(22)

466(6.9)

35(0.5)

41(0.6)

99(1.5)

31(0.5)

78(1.2)

21(0.3)

7(0.1)

67(1.0)

11(0.2)

2(0.03)

43(0.6)

59(0.9)

*Other:4

pseudarth

rosis,1CN

XIIinjury,1pneumonia,2

cerebrospinalfluidleaks,1urinarytra

ctinfection,

1urinaryretention.

yOther:1

ileus,1

seroma.

zOther:1

wound

dehiscence,3

retained

drain,1pleuraleffu

sion,1

cardiacarrhythm

ia;1

appendectomy,2superiorm

esenteric

arterysyndrome,1myocardialinfarction,1postoperativeosteom

yelitis,2

cerebrospinalfluidleak,1

aorticinjury,2

wound

necrosisrequiring

revision,3

ileus,2

urinarytra

ctinfection,1ulnarn

europathyfro

mpositioning,1

stroke,1

brachialplexopathy,1

postoperativehematom

arequiring

reoperation,1intercostalarte

ryinjuryrequiring

reoperation,1

pseudarth

rosis,1seroma,

1peroneal

nervepalsyfro

mpositioning,1

renalveinthrombosis.

PEER-REVIEW REPORTS

BRIAN J. WILLIAMS ET AL. COMPLICATIONS OF TRAUMATIC SPINAL FRACTURE SURGERY

ComplicationsThe overall incidence of complicationswas 6.9% (466 of 6706). A detailed list ofoverall complications and complicationsstratified by fracture level is shown inTable 2. Complications were mostcommon among the occipitocervicalgroup (10%). The perioperative mortalityrate was 0.5% (n ¼ 35). The most commoncause of death was due to a respiratoryetiology (N ¼ 17, 49%), primarilycomprised of respiratory failure (N ¼ 8),followed by a cardiac etiology (N ¼ 6,17%) then sepsis (N ¼ 5, 14%). See Table 3for a detailed analysis of all mortalities inthis series. Age was not significantlyassociated with perioperative mortalitywhen analyzed as 2 groups: �60 or <60(P ¼ .99) or as a continuous variable (P ¼.83). There was no significant differencebetween the median age of patientswho developed complications, 59 years(1 to 95), compared to those who did not,61 years (1 to 97, P ¼ .27). When groupedinto age �60 or <60, there was nota significant difference in the incidence ofcomplications (P ¼ .29) (Table 4). Whenanalyzed as a continuous variable, therewas no significant correlation betweencomplications and age (P ¼ .3). Therewere significantly more complications inthe occipitocervical group compared tothoracolumbar (10% vs. 6.7%, P ¼ .003);however, there was no significant differ-ence between occipitocervical and subaxialcervical groups (10% vs. 7.9%, P ¼ .14) orsubaxial cervical and thoracolumbar (7.9%vs. 6.7%, P ¼ .13). There were significantlymore complications in patients treatedwith spinal fusion (n ¼ 314, 8.7% vs.n ¼ 153, 5%; P < .001). The combined(anterior/posterior) approach fusions hadthe highest complication rate, 10.8%,whereas the anterior-only approachhad the lowest, 6.8%. The combinedapproaches had significantly more com-plications than the anterior approach (P ¼.02); however, neither (combined noranterior) were significantly different fromthe posterior approach (P ¼ .19 andP ¼ .48, respectively) (Table 5).For the entire database population, the

incidence of overall complications was7.5% [(8093 of 108,478), P ¼ .16],mortality was 0.2% [(198 of 108,478),P < .0001], and new postoperativeneurological deficit was 1% [(1064 of108,478), P ¼ .4]. Only the incidence of

WORLD NEUROSURGERY- (-): ---, MONTH 2014 www.WORLDNEUROSURGERY.org 3

Table 3. Detailed Analysis of the Cause of Death for All Mortalities from 6706 Patientswith Traumatic Spinal Fractures

Cause of Death N Percent of Overall Morality (%)

Respiratory 17 49

Respiratory failure 8

Pneumonia 1

Suspected pulmonary embolism 1

Pulmonary embolism 5

Acute respiratory distress syndrome 1

Aspiration 1

Cardiac 6 17

Arrest 4

Arrhythmia 1

Unclear/right ventricular hypertrophy/dilated right ventricle 1

Sepsis 5 14

Stroke 2 6

Hemorrhagic cerebellar 1

Ischemic brainstem 1

Tension pneumothorax* 1 3

Gastrointestinal bleed 2 6

Esophageal varices 1

Lower gastrointestinal 1

Intraoperative blood loss 1 3

Not specified 1 3

Total 35

*Central line malposition preoperatively.

Table 4. Incidence of ComplicationsStratified by Age for 6706 Patients withTraumatic Spinal Fracture

Age(years)

Total(N)

Complications(N) Rate

0e19 540 32 5.9%

20e39 1196 79 6.6%

40e59 766 65 8.5%

50e59 743 60 8.1%

60e69 897 64 7.1%*

70e79 1325 77 5.8%

80e89 1064 75 7.0%

>90 132 8 6.1%

Notrecorded

43 6 14.0%

Overall 6706 466 6.9%

*When analyzed as age >60 or <60, there was nosignificant difference in the incidence of complica-tions (P ¼ .29). When analyzed as a continuousvariable, there was no correlation between age andcomplications (P ¼ .3).

PEER-REVIEW REPORTS

BRIAN J. WILLIAMS ET AL. COMPLICATIONS OF TRAUMATIC SPINAL FRACTURE SURGERY

mortality was significantly different whencomparing the spinal fracture group(0.5%) to the remaining population of thedatabase (0.2%, P < .0001).

New Neurological DeficitsThere were 59 (0.9%) new postoperativeneurological deficits. See Figure 1 fora detailed description of their time course,type of injury and recovery pattern. Ingeneral, new postoperative deficits tendedto recover at least partially. The exceptionswere complete cord injuries, which tendedto remain stable. There were significantlymore new neurological deficits associatedwith occipitocervical (1.8%) and subaxialcervical spine (1.4%) surgery comparedwith thoracolumbar (0.7%, P ¼ .03 and.04, respectively). Use of neuromonitoringwas associated with development ofnew postoperative neurological deficits

4 www.SCIENCEDIRECT.com

(n ¼ 19, 0.5% vs. n ¼ 44, 1.6%; P < .001),likely reflective of an association betweencase complexity and the decision to useneuromonitoring. There was no signifi-cant difference in new postoperativeneurological deficits between patients whounderwent a fusion compared to thosewho did not (23 of 3076, 0.7% vs. 40 of3619, 1.1%; P ¼ .3). The incidence of newneurological deficit was significantlyhigher when a dural tear was notedintraoperatively, 4 of 78 (5.1%), comparedto cases without a dural tear, 55 of 6613(0.8%, P < .0001). The incidence of dur-otomy was not higher among patients witha preoperative neurological deficit, 16 of1452 (1.1%), compared to those without,62 of 4591 (1.35%, P ¼ .4). The rate of newneurological deficit did not differ signifi-cantly among the various approaches forfusion (P ¼ .6) (Table 5).

WORLD NEUROSURGERY, http://

Predictors of Complications and NewNeurological DeficitsUnivariate analysis demonstrated occipito-cervical fracture level (P ¼ .001), presenceof preoperative neurological deficit (P ¼.001), and fusion (P< .001) as predictors ofcomplications. Of note, the fusion groupwas younger (45� 14 vs. 70� 16, P< .001),more likely to have a preoperative neuro-logical deficit (1338 [37%] vs. 114 [3.7%],P < .001), and more likely to have neuro-monitoring (2247 [62%] vs. 1697 [55%],P < .001) and a different fracture regionthan nonfusion cases (occipitocervical 299[8.3%] vs. 100 [3.2%], P < .001; subaxialcervical 863 [24%] vs. 47 [1.5%], P < .001;thoracolumbar 2422 [67%] vs. 2860[92.5%], P < .001). On univariate analysis,fracture level (P ¼ .01), preoperativeneurological deficit (P ¼ .002), and use ofintraoperative neuromonitoring (P < .001)were predictors of new neurological deficit.On multivariate analysis, preoperativeneurological deficit (P ¼ .001; odds ratio[OR] 1.449, 95% confidence interval [CI]1.156 to 1.817) and fusion (P¼ .001;OR 1.12,95% CI 1.072 to 1.168) were both predictorsof complications; however, fracture levelwas not part of the best fit model. Onmultivariate analysis, use of intraoperative

dx.doi.org/10.1016/j.wneu.2013.02.013

Table 5. Incidence of Complications and New Neurological Deficits for 3617 PatientsSurgically Treated with Traumatic Spinal Fractures Stratified by Approach and FusionMethod

Approach Total Cases (N)Complications

(N, %)New NeurologicalDeficits (N, %)

Posterior 2489 220 (8.8)* 29 (1.2)

Intralaminar 1132 87 (7.6) 12 (1.1)

Posterolateral 1253 117 (9.4) 16 (1.3)

Transforaminal lumbar interbody fusion 53 10 (19) 0

Posterior lumbar interbody fusion 51 6 (12) 1 (2)

Anterior 659 45 (6.8)* 4 (0.6)

Combined 427 46 (10.8)* 7 (1.6)

Anterior/posterior 412 45 (10.9) 7 (1.7)

Posterior/anterior/posterior 15 1 (6.7) 0

Not recorded 44 3 (6.8) 0

Overall 3617 314 (8.7) 40 (1.1)

*Combined approaches had significantly more complications compared to anterior approaches (P ¼ .02), but not more thanposterior approaches (P ¼ .19).

PEER-REVIEW REPORTS

BRIAN J. WILLIAMS ET AL. COMPLICATIONS OF TRAUMATIC SPINAL FRACTURE SURGERY

neuromonitoring (P¼ .016; OR 1.949, 95%CI 1.13 to 3.361) and preoperative neuro-logical deficit (P < .001; OR 2.964, 95% CI1.667 to 5.271) were predictors of newneurological deficits; however, fracturelevel and durotomy were not a part of thebest fit model.

DISCUSSION

We present a large series of surgicallytreated spinal fractures focusing oncomplications related to 6706 cases.Twenty-two percent of patients hada preoperative neurological deficit, andsubaxial cervical spinal fractures were mostlikely to have a preoperative deficit, 43%.The overall incidence of complications was6.9%, and overall mortality was 0.5%.Fusion was performed in 54% of cases.Factors found to be predictive of complica-tions on univariate analysis includedperforming a fusion, fracture level, andpreoperative neurological deficit. Onmultivariate analysis, both preoperativeneurological deficit and fusion werepredictors of complications. These data aresupported by a recent study of a traumadatabase that included 230 patients andfound that use of high-dose steroids,American Spinal Injury Association score,

WORLD NEUROSURGERY- (-): ---, MON

and Charlson Comorbidity Index werepredictors of complications (4).The incidence of complications reported

in the literature varies widely depending onthe series, which is comprised of variousfracture types and patient characteristics.Reinhold et al. in a prospective observa-tional trial noted a 15% incidence of overallcomplications and 1.2% mortality in anevaluation of 773 cases of surgically treatedthoracolumbar fractures (15). They noted 56(7.7%) occurred intraoperatively; however,they included conversion from endoscopicto an open technique as a complicationalong with hemorrhage and instrumenta-tion malposition. With the 8 cases ofconversion from an endoscopic techniqueremoved, the intraoperative complicationrate was 6.5% (n ¼ 48). The postoperativecomplication rate was 9.4% (n ¼ 69), withthe most common complications in themiscellaneous category (n ¼ 24), thenwound healing problems other than infec-tion (n ¼ 14). They noted a significantlyhigher incidence of complication amongthe anterior/posterior group (10.7%,n ¼ 34) compared to the posterior alonegroup (5.9%, n ¼ 22). They did notcomment on the relationship betweenneurological function and the incidence ofperioperative complications. Their overallcomplication rate is higher than our noted

TH 2014

overall rate of complications, which is likelydue to differences in spinal injuries,comorbid injuries, and method of treat-ment (15). Nasser et al. performeda comprehensive, systematic review ofcomplications related to all spine surgeryand found that the overall incidence ofcomplications for all spinal surgery was16.4%. They noted a higher incidence ofcomplications in prospective studies andstudies with longer follow-up. They did notfind an association between approach inany region of the spine and incidence ofcomplications. Importantly, they note thesignificant variability regarding the defini-tion of a complication compared to anadverse event. This variability undoubtedlycontributes to the heterogeneity found inthe literature and a part of the differencebetween our study and some of the existingliterature (13).In support of our lower complication

rate, Rampersaud et al. published theresults of a prospective observational studyfocused on perioperative complicationsfrom all spinal surgeries (n ¼ 700). Theyreported an intraoperative adverse event(unexpected or undesirable event) rate of14% (n ¼ 98); however, only 23 of98 (23.5% or 2.3% overall) of those adverseevents resulted in a clinically relevantsequelae or complication. However, theirstudy was comprised of only 9.6% trau-matic pathology (14). Regarding a trauma-specific population of patients, Verlaanet al. performed a systematic review of theliterature for surgical treatment of trau-matic spinal injuries with a specificfocus on the relevance of approach. Theoverall incidence of complications fromall approaches was 6.6% (283 of 4304recorded), a similar rate to our study,consistent with our reported incidence ofoverall complications (21).The overall incidence of newneurological

deficits in our study was 0.9%. The mostlikely level of injury to be associated witha new deficit was the occipitocervical level,1.8%, followed closely by the subaxialcervical spine, 1.4%. In general, newneurological deficits occurred in the acuteperiod after surgery (<24 hours), andnerve root deficits were the most commontype. All deficits tended to improve at leastpartially regardless of onset or type.Complete cord deficits were the exception,and typically remained complete. Otherseries have reported similar findings of

www.WORLDNEUROSURGERY.org 5

Total Number ofInjuries Timing of Injury Type of lnjury

Recovery fromInjury

Recovery legendNone = norecovery

recovery

recovery

Partial = some

Full = complete

Not RecordedN=4

IntraoperativeN=10

Acute(<24 hours)

N=29

Delayed(>24hours)

*The type and recoveryof two injuries were notrecorded

New NeurologicalDeficitsN=59

N=16*

Nerve root, N=2

Complete cord, N=1

Incomplete cord,N=7

Nerve root, N=19

Complete cord, N=1

Incomplete cord,N=8

Cauda Equina, N=1

Partial, N=1None, N=1

None, N=1

Full, N=2Partial, N=4None, N=1

None, N=1

Full, N=5Partial, N=14

None, N=1

Full, N=4

Partial, N=1

Partial, N=2Nerve root, N=10

Complete cord, N=1

Incomplete cord,N=1

Cauda Equina, N=1

Partial, N=1

Partial, N=1Full, N=1

Full, N=4Partial, N=4

Figure 1. New neurological deficits after surgery for 6706 cases of spinal fractures stratified by timecourse, type of injury, and recovery. *The type and recovery of 2 injuries were not recorded.

PEER-REVIEW REPORTS

BRIAN J. WILLIAMS ET AL. COMPLICATIONS OF TRAUMATIC SPINAL FRACTURE SURGERY

neurological injury (8, 21). Use of neuro-monitoring was associated with newneurological deficits on univariate andmultivariate analysis. However, we believethat this likely reflects a group of patient-s that are at higher risk for neurologicaldeficit preoperatively. We believe that thesurgeon was more likely to use neuro-monitoring on cases that were morecomplex, had longer expected operativetimes, and had more concern for potentialof new neurological deficit associated withthe operative procedure. In addition,preoperative neurological deficit wasa predictor of new neurological deficit.Similarly, we believe that the increasedincidence of complications associated withfusion for treatment of these injuries likelyreflects at least in part the complexity of theprocedure and fracture, rather than a trueincreased risk directly introduced by thefusion procedure.There are several limitations to the

present study. This is a retrospective studyand is subject to the inherent weaknesses ofthis methodology. Some key informationnot included in the database includesdocumentation of medical comorbidities,timing of surgery, American Spinal InjuryAssociation grade, associated significantinjuries, case complexity, and long-term

6 www.SCIENCEDIRECT.com

follow-up. Unfortunately the database isdeidentified, and we are unable to obtainfurther information to augment the currentdata. Given that this database was designedfor the assessment of perioperativecomplications and the data were input bythe surgeons, it is possible that minor,distant, or seemingly unrelated medicalcomplications were underreported. Theinclusion of only perioperative complica-tions also does not accurately reflect thetrue incidence of pseudarthrosis or thelong-term recovery from neurologicalinjury. Finally, we must reiterate that thesecases are self-reported by members of theSRS. Candidate members were required tosubmit their MM data, whereas activemembers were encouraged to submit theirdata. Due to these limitations of this data-base, we must conclude that the compli-cation rates represent an estimate on thelow end for these procedures. There iscurrently no method to determine thecompleteness of data submission, nor theaccuracy of the reporting.

CONCLUSIONS

Overall, surgery for the treatment of spinalfractures was performedwith a relatively lowincidence of perioperative complications

WORLD NEUROSURGERY, http://

(6.9%) and mortality (0.5%). Respiratoryetiology was the most common cause ofmortality. The occipitocervical region wasmost likely to be associated with complica-tions (10%), mortality (1.5%), and newneurological deficits (1.8%). Preoperativeneurological deficit anduse of intraoperativeneuromonitoring (likely reflective of casecomplexity) were predictors of new neuro-logical deficits, and preoperative neurolog-ical deficit and fusion were predictors ofcomplications. These data may prove usefulfor patient counseling andongoing efforts toimprove the safety of operative care forpatients with spinal fractures.

REFERENCES

1. Anderson PA, Henley MB, Rivara FP, Maier RV:Flexion distraction and chance injuries to thethoracolumbar spine. J Orthop Trauma 5:153-160,1991.

2. Cengiz SL, Kalkan E, Bayir A, Ilik K, Basefer A:Timing of thoracolomber spine stabilization intrauma patients; impact on neurological outcomeand clinical course. A real prospective (RCT)randomized controlled study. Arch OrthopTrauma Surg 128:959-966, 2008.

3. Chipman JG, Deuser WE, Beilman GJ: Earlysurgery for thoracolumbar spine injuries decreasescomplications. J Trauma 56:52-57, 2004.

4. Dimar JR, Fisher C, Vaccaro AR, Okonkwo DO,Dvorak M, Fehlings M, Rampersaud R,Carreon LY: Predictors of complications afterspinal stabilization of thoracolumbar spineinjuries. J Trauma 69:1497-1500, 2010.

5. Jansson KA, Blomqvist P, Svedmark P, Granath F,Buskens E, Larsson M, Adami J: Thoracolumbarvertebral fractures in Sweden: an analysis of13,496 patients admitted to hospital. Eur J Epi-demiol 25:431-437, 2010.

6. Leucht P, Fischer K, Muhr G, Mueller EJ: Epide-miology of traumatic spine fractures. Injury 40:166-172, 2009.

7. Levi L, Wolf A, Rigamonti D, Ragheb J, Mirvis S,Robinson WL: Anterior decompression in cervicalspine trauma: does the timing of surgery affect theoutcome? Neurosurgery 29:216-222, 1991.

8. Marre B, Ballesteros V, Martinez C, Zamorano JJ,Ilabaca F, Munjin M, Yurac R, Urzua A,Lecaros M, Fleiderman J: Thoracic spine fractures:injury profile and outcomes of a surgically treatedcohort. Eur Spine J 20:1427-1433, 2011.

9. McHenry TP, Mirza SK, Wang J, Wade CE,O’Keefe GE, Dailey AT, Schreiber MA,Chapman JR: Risk factors for respiratory failurefollowing operative stabilization of thoracic andlumbar spine fractures. J Bone Joint Surg Am 88:997-1005, 2006.

10. McLain RF, Benson DR: Urgent surgical stabili-zation of spinal fractures in polytrauma patients.Spine (Phila Pa 1976) 24:1646-1654, 1999.

dx.doi.org/10.1016/j.wneu.2013.02.013

PEER-REVIEW REPORTS

BRIAN J. WILLIAMS ET AL. COMPLICATIONS OF TRAUMATIC SPINAL FRACTURE SURGERY

11. Mithani SK, St-Hilaire H, Brooke BS, Smith IM,Bluebond-Langner R, Rodriguez ED: Predictablepatterns of intracranial and cervical spine injuryin craniomaxillofacial trauma: analysis of 4786patients. Plast Reconstr Surg 123:1293-1301,2009.

12. Mulligan RP, Friedman JA, Mahabir RC:A nationwide review of the associations amongcervical spine injuries, head injuries, and facialfractures. J Trauma 68:587-592, 2010.

13. Nasser R, Yadla S, Maltenfort MG, Harrop JS,Anderson DG, Vaccaro AR, Sharan AD, Ratliff JK:Complications in spine surgery. J Neurosurg Spine13:144-157, 2010.

14. Rampersaud YR, Moro ER, Neary MA, White K,Lewis SJ, Massicotte EM, Fehlings MG: Intra-operative adverse events and related postoperativecomplications in spine surgery: implications forenhancing patient safety founded on evidence-based protocols. Spine (Phila Pa 1976) 31:1503-1510, 2006.

15. Reinhold M, Knop C, Beisse R, Audige L,Kandziora F, Pizanis A, Pranzl R, Gercek E,Schultheiss M, Weckbach A, Buhren V, Blauth M:Operative treatment of 733 patients with acutethoracolumbar spinal injuries: comprehensiveresults from the second, prospective, Internet-based multicenter study of the Spine Study Groupof the German Association of Trauma Surgery. EurSpine J 19:1657-1676, 2010.

16. Robertson A, Branfoot T, Barlow IF,Giannoudis PV: Spinal injury patterns resultingfrom car and motorcycle accidents. Spine (PhilaPa 1976) 27:2825-2830, 2002.

17. Robertson A, Giannoudis PV, Branfoot T,Barlow I, Matthews SJ, Smith RM: Spinal injuries

WORLD NEUROSURGERY- (-): ---, MON

in motorcycle crashes: patterns and outcomes.J Trauma 53:5-8, 2002.

18. Sapkas G, Kateros K, Papadakis SA, Galanakos S,Brilakis E, Machairas G, Katonis P: Surgicaloutcome after spinal fractures in patients withankylosing spondylitis. BMC Musculoskelet Dis-ord 10:96, 2009.

19. Schinkel C, Frangen TM, Kmetic A, Andress HJ,Muhr G: Timing of thoracic spine stabilization intrauma patients: impact on clinical course andoutcome. J Trauma 61:156-160, 2006 [discussion160].

20. Smith JS, Fu KM, Polly DW Jr, Sansur CA,Berven SH, Broadstone PA, Choma TJ, Goytan MJ,Noordeen HH, Knapp DR Jr, Hart RA,Donaldson WF 3rd, Perra JH, Boachie-Adjei O,Shaffrey CI: Complication rates of three commonspine procedures and rates of thromboembolismfollowing spine surgery based on 108,419 proce-dures: a report from the Scoliosis ResearchSociety Morbidity and Mortality Committee. Spine(Phila Pa 1976) 35:2140-2149, 2010.

21. Verlaan JJ,DiekerhofCH,BuskensE, vanderTweel I,VerboutAJ,DhertWJ,Oner FC: Surgical treatmentoftraumatic fractures of the thoracic and lumbar spine:a systematic review of the literature on techniques,complications, and outcome. Spine (Phila Pa 1976)29:803-814, 2004.

Conflict of interest statement: J. Smith is a consultant forDepuy, Biomet, Medtronic, and a former consultant for AxialBiotech; has received payment for lectures from DePuy,Biomet, Medtronic, and Globus; and has performed researchstudy group support for The Spinal Deformity Study Groupand International Spine Study Group for DePuy andMedtronic. C. Ames is a consultant for DePuy, Stryker, andMedtronic; has received a Trans 1 grant in the past; has held

TH 2014

patents from Fish & Richardson, P.C., in the past; andreceives royalties from Aesculap and Lawx. L. Lenke receivesroyalties from Medtronic and Quality Medical Publishing,and receives research support from Depuy Spine and AxialBiotech. P. Broadstone has received payment for lecturesfrom Synthes and Medtronic, and is a stock holder ofPhygen. A. Vaccaro is a consultant for Gerson LehrmanGroup, Guidepoint Global, and Medacorp; has receivedgrants from Stryker Spine and Cerapedics; receives royaltiesfrom DePuy, Medtronics, Stryker Spine, Biomet Spine,Globus, Aesculap, and Nuvasive; and is a stock holder ofReplication Medica, Globus, K-2 Medical, Paradigm Spine,Stout Medical, Spine Medica, Computational Biodynamics,Progressive Spinal Technologies, Spinology, Orthovita,Vertiflex, Small Bone Innovations, Disk Motion Technology,NeuCore, Cross Current, Syndicom, In Vivo, Flagship Surgical,Advanced Spinal Intellectual Properties, Cytonics, BonovoOrthopaedics, Electrolux, Gamma Spine, Location BasedIntelligence, FlowPharma, and R.I.S. D. Polly Jr. is a formerconsultant for Medtronic and a board member of theScoliosis Research Society (travel expenses to somemeetings). C. Shaffrey is a consultant for Biomet, Depuy, andMedtronic; a board member of the American Board ofNeurological Surgery, American Association of NeurologicalSurgery, and Scoliosis Research Society; has received grantsfrom the National Institutes of Health, Department ofDefense, North American Clinical Trials Network, and AO;holds patents with Medtronic and Biomet; received royaltiesfrom Medtronic; has received payment for lectures fromStryker; and is an Editorial Board Member of the Journal ofNeurosurgery, Neurosurgery, and Spine. B. Williams and D.Saulle have no conflicts of interest to report.

Received 13 August 2012; accepted 2 February 2013

Citation: World Neurosurg. (2014).http://dx.doi.org/10.1016/j.wneu.2013.02.013

Journal homepage: www.WORLDNEUROSURGERY.org

Available online: www.sciencedirect.com

1878-8750/$ - see front matter ª 2014 Elsevier Inc.All rights reserved.

www.WORLDNEUROSURGERY.org 7