Spine
SPINE Volume 39, Number 11, pp 911-918 ©2014, Lippincott Williams &c Wilkins
SURGERY
Venous Thromboembolism After Spine Surgery William W. Schairer, MD,* Andrew C. Pecltke, MD,t and Serena S. Hu, MD+
Study Design. Retrospective cohort study. Objective. To measure the rate of postoperative venous thromboembolic events (VTE) after spine decompression and fusion procedures. Summary of Background Data. VTE after spine surgery is a serious complication, but chemoprophylaxis is not without significant risk due to the concern of epidural hematoma. Current literature report widely variable rates of VTE, and have weaknesses in sample size, specificity of diagnosis, and methodological problems with adequate patient follow-up. Methods. State-level inpatient, ambulatory surgery, and emergency department administrative databases were used to track patients for clinically significant VTE within 90 days of discharge after a spine procedure. Results. Of 357,926 patients enrolled, one-third underwent spine decompression alone, whereas two-thirds received a spine fusion. The overall rate of VTE was 1.37% (95% Cl: 1.33-1.41 ), but varied widely depending on diagnosis, 1.03% for structural degenerative diagnoses to 10.7% for spine infection. Posterior cervical fusion had a higher rate of VTE than anterior cervical fusion, whereas anterior thoracolumbar and lumbosacral fusions had higher rates than the respective posterior approaches. Additional risk factors included patients receiving long spine fusions and having multiple procedures during the hospitalization. Eorty percent of VTEs discovered after discharge were diagnosed at a different hospital. Conclusion. The rate of spine VTE varies widely depending on diagnosis and procedure. It is important to risk-stratify patients who present for spine surgery to identify patients at increased risk who should be monitored for the development of VTE. It is important to know that nearly half of VTEs that occur after discharge are diagnosed at different hospitals, and thus the primary surgeon may From the *Department of Orthopaedic Surgery, Hospital for Special Surgery, New York, NY; tDepartment of Orthopaedic Surgery, University of California San Francisco, San Francisco, CA; and tDepartment of Orthopaedic Surgery, Stanford University School of Medicine, Redwood City, CA. Acknowledgment date: Septemher 16, 2013. First revision date: December 8, 2013. Second revision date: February 1, 2014. Third revision date: February 23, 2014. Acceptance date: February 25, 2014. The manuscript submitted does not contain information about medical device(s)/drug(s). No funds were received in support of this work. Relevant financial activities outside the submitted work: consultancy, payment for lecture, royalties, stocks, travel/accommodations/meeting expenses. Address correspondence and reprint requests to Serena S. FHu, MD, Department of Orthopaedic Surgery, University of California San Francisco, 500 Parnassus Ave, M U 320W, San Francisco, CA 94143-0728; F-mail: HuS@ orthosurg.ucsf.edu DOI: 10.1097/BRS.0000000000000315 Spine
be initially unaware of the complication. These results from a large selection of historical patients may provide a tool for estimating patient risk depending on diagnosis and type of procedure. Key words: spine surgery, decompression, fusion, deep venous thrombosis, venous thromboembolic event, pulmonary embolism, PE, DVT, VTE, complication. Level of Evidence: 2 Spine 2014;39:911-918
V
enous thromboembolism (VTE), which includes deep venous thrombosis (DVT) and pulmonary embolism (PE), are serious complications after surgery. Although in general there are consensus guidelines about early postoperative pharmacological prophylaxis after most surgical procedures, there is less agreement about VTE prophylaxis methods after spine surgery. In addition to an increased risk of infection, anticoagulation leads to increased venous bleeding and increases the risk of a potentially catastrophic neurological complication due to formation of an epidural hematoma compressing the spinal cord.'"^ In addition, aspirin, even when discontinued a week preoperatively, has been shown to cause increased blood drainage and increased transfusion requirements.'' Not surprisingly, there is a wide variability in practice among spine surgeons regarding the use and timing of VTE prophylaxis," as well as wide variation across hospitals.** In the literature, there is a wide range of reported rates of VTE after spine surgery, from as low as 0.29% and up to 23.4%.'*"'^ This disparity is due to a number of factors. One reason is the different rates reported in patients with different indications for spine surgery. In addition, some studies define VTE as only those that are detected clinically, whereas others screened all patients to detect subclinical events. A major methodological flaw is the lack of appropriate follow-up, given that patients seeking urgent and emergent care may not present to the hospital where they underwent surgery. Einally, different time periods for follow-up may significantly impact the number of events detected. Although VTEs are a well-known complication, accurate assessment of the prevalence is of the utmost importance in patients who undergo spine surgery. The American College of Chest Physicians state that patients with risk factors who are undergoing elective spine surgery should receive postoperative low-dose unfractionated heparin or low-molecular weight heparin." However, in their consensus clinical guidelines for antithrombotic therapy in spine surgery. The North American Spine Society was unable to agree on the rate of www.spinejournal.com
911
SURGERY
clinically significant VTE due to the limitations in published data.^ One must weigh the risks and benefits of postoperative chemoprophylaxis, and the risks may be very different depending on the type of procedure and individual patient diagnosis. As such, the North American Spine Society clinical guidelines state that there are not enough high-quality studies with subgroup analyses that allow identifying the risk in different patient groups. Thus, this study set out to address the shortcomings in the current literature regarding the rate of VTF after spine surgery. This study was designed to use a large patient cohort that would allow appropriate subgroup analyses for patients with different diagnoses and procedures. In addition, patient follow-up is the key to accurate diagnoses; this study allowed us to track patients to any hospital within their home state to identify clinically significant VTE up to 6 months after spine surgery. In addition, the secondary purpose of this study was to identify risk factors for VTE depending on patient characteristics and procedure details.
Venous Thromboembolism After Spine Surgery • Schairer et al
Data Sources SID Database
California, Fiorida 2005 - 2009
SASD Database
TT Spine Decompression or Fusion
Group Seiection
387,924 patients Insufficient Follow-up: 21,653 patients
Out-of-state: 8345 patients 357,926 patients
Outcome Measurement
SID Database
/
\ \
SEDD Database
/ 90-day follow-up
Inpatient or ED diagnosis of VTE Figure 1 . Data sources from California and Florida: 2005-2009.
MATERIALS AN D METHODS Data Sources We used the State Inpatient Database (SID), State Ambulatory Surgery Database (SASD), and the State Emergency Department Database (SFDD). These administrative claims databases are maintained by the Healthcare Cost and Utilization Project, which is part of the Agency for Healthcare Research and Quahty. The databases contain 100% of patient records for each available state. Each visit contains information on patient demographics. International Classification of Diseases, Ninth Revision (ICD-9) diagnosis codes, and ICD-9 and Current Procedure Terminology codes (Current Procedure Terminology only in SASD, 7CD-9 in SASD and SID). We compiled the SID, SASD, and SEDD databases from California (2005-2009) and Florida (2005-2009). These states and years were chosen because they participated in all 3 databases (SID, SASD, and SEDD) and they contained a unique identifier to track patient visits between databases and over time. This ensured that we could identify patients with clinically significant VTE whether they returned to the same hospital or went to a different hospital (Figure 1).
Patient Selection Procedure codes were used to identify patients who underwent spine fusion and nonfusion procedures either as an inpatient (in the SID) or outpatient (in the SASD). Patients were excluded if their home state was different than the state of the hospital where they underwent their spine surgery. This was done because in an acute setting, these patients may first seek medical care in their home state and would thus not be captured in our data. In addition, patients with less than 90 days of available follow-up (spine surgery in October, November, or December of 2009) were excluded. Surgical procedures were identified as cervical, thoracolumbar, or lumbosacral. Patients were grouped by 1 of 5 primary diagnoses: structural (spondylosis, spondylolisthesis, 912
www.spinejournal.com
scoliosis, other), complication of implant or previous surgery, trauma (including vertebral compression fractures), infection, and cancer. Medical risk factors associated with VTE were identified using ICD-9 diagnosis codes, including some groups from the Elixhauser medical comorbidity definitions.''' These risk factors were included in a multivariate model to increase the specificity of risk associated with each diagnosis and procedure.
Identification of Venous Thromboembolic Events Clinically relevant DVT and PE were identified by ICD-9 diagnosis codes. Events occurred during the index hospital visit (SID or SASD databases) or within 90 days after discharge (SID or SEDD databases). To isolate VTE risk related to spine surgery, patients who were readmitted to the hospital or presented to the emergency department were censored from that point forward. In addition, patients were censored after 1 VTE event to avoid counting patients more than once.
Statistical Analysis
The incidence of VTE occurring within 90 days of discharge was evaluated using a Kaplan-Meier time-to-event analysis; groups were compared with a log-rank test. Event rates are presented both as the Kaplan-Meier estimation and as the absolute number of events. Risk factors were evaluated using a Cox proportional hazards model, and included patient demographics, surgical procedure, and diagnosis. To better evaluate the different spine fusion procedures, each was compared with decompression alone. In addition, multivariate models were stratified by medical comorbidities related to adjust for different baseline risk for each patient (history of DVT, atrial fibrillation, use of anticoagulation medication, coagulation deficiency, current tobacco use, peripheral vascular disease, pregnancy, obesity, coronary artery disease, lymphoma, and cancer). May 2014
Spine
SURGERY
Venous Thromboembolism After Spine Surgery • Schairer et al
C O
U
^
LO
ro
O
o
ro O
O
rM
O
LO
^
I—
LO
ro O
CN
CO
ro
CN
LO
CO
CO
LO
tn
IX
tn
o
tN
LO
CO
tN
O
o
O
evic Sur mpl Itio
s? ai .w
o Q c
LO
CO
I—
O
ro
tn O
CO
o
O CN
tN
r-
tN
o "^
CN
LO
ro
O
ro
CO
tN
IX
O
O
ro
D Uo 01
IX
E 5^ o
tn
CO CO
tN
ro
CO
CO
O
U
¡^
•
V.D
*"
3 ^^
ro
LO
t í S?
,—
tn
LO CTl
O
O
LO
c
O
en
CO
CO tN
Ix .^ 0^
CO CN
^
ro O
ro
tN
ro
LO
LO
rN
IX
tN
LO
m
^ tN
Ix ,
Ix
tN
tN
tN
CO tN
ro
ro ro
CN
tN
IX ,
ro
ro
LO
CTi CO
LO
O^
Oi
O
LO
CM
ro
LO
o '-D IX
tN
d
LO
LO
o^
^
Ix
LO
LO
tji
O CN
IX
CO
O
ro
CN ,—
en
o
CN IX
O
o
CO
t-M
ix'
O
LO
rM
O
,
CN
tN
tN
o
O
O
O
O
LO
ro
o
tJi
LO
LO
>—
r—
tN
LO
ro
tN
ro ro
rM
CO LO
ro
CN
tN
(Jl
tN tN
en
00 IX
u O
o
5, lumbc
•s ;""
t3^
4.6
ro
IX -—
r—
2.4
os
r—
MuH Stage
L
^
16.3
o
12.3
S 's
8.3
ity
3
IX
ro
O
5 Ix
Ix
ro
CN
IX
CTi
en
ro
•~O
LO
ro
CN
LO
O
°^ E 5^
ro O
ixi
tn
tTi
LO
ro
CN LO
IX tN LO
^
CO
ro "*
, LO
CO ^
LO
CN
O
CO
CN LO
tTi
LO
.^
"*
CO
tn
CO
IX
CN LO
LO LO
tN LO
LO
fN
CN CO Ü1
LO
tn
tN
ro CO
iri
v£)
tTi
tji LO
CN LO
CO CM
K tN
ro
CN"
ro ro
ro O tn
ro ro LO
CN
nterio r-posti rior
ro
LO
(Ti
a."
LS posterior i usion 1
o
LS AP fusion
LS anter 1
sion
LS déco Tipre ssion 1
rz
Lumbosac
TL posterior 1 usion 1
TL AP fusion
TL anter 1
sion
TL décompre ssion
o
1
Cervica
Q.
n]ïïK-. 1ndicating Diagnosis Structural
Trauma
Cancer
Device/Surgical Complication
Infection
Cervical decompression
0.8 (0.7-1.0)
4.8(1.8-12.2)
8.9(5.3-14.6)
8.4(4.8-14.4)
4.5(1.7-11.7)
Atlas-axis fusion
1.8(1.1-2.9)
1.5(0.2-10.3)
5.5(4.1-7.4)
No events
2.9(0.4-19.1)
Cervical anterior fusion
0.4 (0.3-0.4)
0.7 (0.3-1.6)
3.9(3.3-4.7)
9.8(6.6-14.4)
5.5(3.0-10.2)
Cervical posterior fusion
1.5 (1.3-1.9)
0.6 (0.2-1.8)
5.9(4.8-7.1)
7.7(3.0-19.2]
5.2 (3.1-8.8)
Cervical AP fusion
2.8(2.1-3.6)
2.7 (0.9-8.2)
9.5(7.7-11.7)
6.4(2.3-17.1)
4.7(1.8-11.9)
TL decompression
1.0(0.8-1.3)
1.7(0.4-6.5)
2.3(1.0-5.6)
11.9(7.7-18.3)
2.6 (1.0-6.7)
TL anterior fusion
2.1 (1.2-3.6)
4.8 (0.7-29.3)
6.1 (4.0-9.1)
9.2 (4.7-17.5)
7.3 (4.1-13.1)
TL posterior fusion
0.9(0.7-1.2)
2.4(1.3-4.4)
7.0(6.1-8.1)
7.5(4.3-13.0)
6.5 (4.9-8.6)
TL AP fusion
3.3(2.1-5.1)
5.4(1.4-19.9)
8.2(5.2-12.7)
14.6(9.0-23.0)
10.1 (5.7-17.5)
LS decompression
0.6 (0.5-0.6)
2.3(1.3^.2)
3.1 (2.1^.4)
6.9 (5.0-9.5)
6.1 (3.5-10.6)
LS anterior fusion
1.5 (1.3-1.8)
0.9 (0.3-2.8)
3.6(2.0-6.3)
5.6(2.2-14.3)
6.8(2.5-17.4)
LS posterior fusion
1.1 (1.0-1.1)
1.4 (1.0-1.9)
3.3 (2.6-4.2)
9.1 (5.6-14.5)
5.9(3.8-9.1)
LSAP fusion
1.6(1.4-1.9)
3.6(2.3-5.8)
4.8(2.7-8.5)
11.1 (6.0-20.0)
10.9(5.3-21.8)
2.9 (2.3-3.8)
3.1 (1.4-6.8)
10.1 (7.6-13.3)
11.5(6.2-21.0)
12.7(8.2-19.2)
Procedure
.„^•_xi.r..,
Cervical
Thoracolumbar
Lumbosacral
Cross junctional (cerv/TL, TL/LS)
The vaiues given are Kapian-Meier % (95% confidence intervais). AP indicates anterior-posterior; TL, thoracolumbar; LS, Iumbosacrai; CI, confidence intervai.
RESULTS Demographics We identified 357,926 patients who underwent a spine decompression and/or fusion (Table 1). Thirty-one percent of the procedures were cervical, 6% were thoracolumbar, and 62% were lumbosacral; 1 % of the patients had fusions extending across a junctional level (cervical and thoracolumbar fusion or thoracolumbar and lumbosacral fusion). Decompression procedures made up one-third of procedures, whereas twothirds of procedures included spine fusion. Overall, 2.7% of spine fusions were revision procedures. Structural diagnoses accounted for the vast majority of diagnoses. Trauma accounted for a significant portion of cervical and thoracolumbar diagnoses, and infection and cancer were a prominent diagnosis in the thoracolumbar group. Venous Thromboembolic Events The overall rate of VTE was 1.37% (Kaplan-Meier 95% CI: 1.33-1.41; absolute N = 4657/357,926). DVT occurred in 1.07% of cases (1.04-1.11) (Table 2), whereas PE occurred in 0.45% (0.43-0.47) of cases (Table 3). VTE occurred in 0.92% (0.87-0.98) of decompression cases, 1.10% (1.04-1.17) of cervical fusion cases, 4.54% (4.20-4.91) of thoracolumbar 914
www.spinejournal.com
fusion cases, and 1.74% (1.66-1.82) of lumbosacral fusion cases (Eigure 2). Patients who received longer spine fusion had higher rates of VTE. In short fusions (2-3 vertebrae), 1.29% (1.23-1.34) of patients had a VTE (hazards ratio [HR] = 1.41, ? < 0.001, compared with decompression). Medium (4—8 vertebrae) and long (9 or more vertebrae) spine fusions had VTE rates of 3.05% (2.88-3.23) (HR = 3.45, ? < 0.001) and 2.80% (2.38-3.28) (HR = 3.22, P < 0.001), respectively. Procedures where the patient was in the supine position were associated with a VTE rate of 0.87% (0.81-0.93). However, patients that were in the prone position had a higher rate of 1.55% (1.50-1.60) (HR = 1.77, P < 0.001). There was wide variation in VTE rate depending on diagnosis (Eigure 3). At the time of discharge from the initial hospitalization, the VTE rate of the structural and implant/surgical complication groups was 0.45% (0.42-0.47) and 0.94% (0.74-1.20), respectively. The trauma, infection, and cancer groups were higher at 4.12% (3.81-4.46), 7.72% (6.698.92), and 4.56% (3.88-5.36). At 90 days after discharge, the overall VTE rates were 1.02% (0.99-1.06) and 1.94% (1.63-2.32) for structural and implant/surgical complication, respectively. As with at discharge, the trauma, infection, and cancer groups were higher at 90 days with VTE rates May 2014
SURCERY
"TOLE
Venous Thromboembolism After Spine Surgery • Schairer et al
s^iiisraigryjggjgi^ïTîiJîiirnïïmTirasai^ Indicating Diagnosis Structural
Trauma
Cancer
Device/Surgical Complication
Infection
Cervical decompression
0.4 (0.3-0.6)
1.1 (0.2-7.6)
2.7(1.1-6.3)
5.4(2.6-11.2)
4.9(1.8-12.7)
Atlas-axis fusion
0.5(0.2-1.3)
1.5 (0.2-10.1)
0.9 (0.4-1.9)
No events
2.9(0.4-19.1)
Cervical anterior fusion
0.2 (0.1-0.2)
0.5(0.2-1.2)
1.4(1.0-1.9)
2.3(1.1-5.1)
2.7(1.1-6.6)
Cervical posterior fusion
0.7 (0.5-0.9)
0.2 (0.0-1.4)
1.5(1.0-2.3)
1.9(0.3-12.9)
2.0 (0.8-4.8)
Cervical AP fusion
0.7(0.4-1.2)
1.9(0.5-7.5)
1.5(0.9-2.6)
2.1 (0.3-14.2)
No events
TL decompression
0.5 (0.3-0.7)
No events
2.1 (0.9-5.1)
5.2(2.5-10.6)
No events
TL anterior fusion
1.3(0.6-2.8)
4.8 (0.7-29.3)
1.8(0.8-3.7)
2.3 (0.6-8.8)
1.3 (0.3-5.3)
TL posterior fusion
0.3 (0.2-0.5)
0.5(0.1-2.1)
2.3(1.7-3.0)
4.6(2.3-9.1)
2.7(1.8-4.1)
TLAP fusion
1.5(0.8-3.0)
No events
3.2(1.6-6.4)
2.4 (0.8-7.4)
4.2(1.7-10.2)
LS decompression
0.3 (0.2-0.3)
0.4(0.1-1.7)
0.7(0.3-1.5)
3.1 (1.9-5.0)
3.3(1.5-7.4)
LS anterior fusion
0.5 (0.3-0.7)
No events
2.3(1.2-4.6)
1.4(0.2-9.6)
4.6(1.1-18.4)
LS posterior fusion
0.6 (0.5-0.6)
0.6 (0.4-1.0)
1.0(0.7-1.6)
No events
1.6(0.7-3.7)
LSAP fusion
0.7 (0.6-0.9)
0.7(0.2-2.1)
2.5(1.2-5.6)
No events
7.5(2.8-19.3)
1.5(1.0-2.1)
No events
3.5 (2.1-5.7)
4.7(1.5-14.5)
7.3(3.7-14.1)
Procedure Cervical
Thoracolumbar
Lumbosacral
Cross junctional (cerv/TL, TL/LS)
The values given are Kaplan-Meier pereent (95% eonfidenee intervais). AP indicates anterior-posterior; TL, thoraeolumbar; LS, lumbosacral; eerv, cervical.
of 6.41% (6.01-6.85), 10.67% (9.37-12.15), and 8.26% (7.24-9.41). Timing of Events Overall, 52.7% (n = 2454) of VTEs occurred during the index hospitalization, whereas 47.3% (n = 2203) occurred after
Postoperative Venous Thromboembolism
discharge (median time, 9.1 d; standard deviation 19.6 d). Events diagnosed during the index hospitalization were PEs in 29.6% of cases compared with 35.9% of cases diagnosed after discharge (P < 0.001). The mortality rate of PE diagnosed during the initial hospitalization was 4.0%, whereas it was lower if diagnosed after discharge at 2.6%. In addition, 40% of VTEs found after discharge were diagnosed at a different hospital than where the initial procedure was performed.
By Procedure Type
Multivariate Model
5.0%-I
4.0%-
3,0% H
^ ^ ----• ^— ^ " -
2.0%-
Thoracolumbar fusion Lumbosacral fusion Cervical fusion Decompression
1.0%-
0.030 60 Days since discharge
90
Figure 2. Postoperative venous thromhoembolism by procedure type. Spine
Independent risk factors assessed through a multivariate model that controlled or a number of medical comorbidities that may affect VTE risk. The model showed an increasing risk of VTE with increasing age (Table 4). In addition, male sex was associated with an increased risk compared with female sex. For the type of surgical procedure, compared with decompression alone, all fusion procedures had higher risk of VTE with the exception of atlas-axis fusion. Although anterior cervical fusion had a lower risk than posterior cervical fusion, the anterior thoracolumbar and lumbosacral fusions had greater risk of VTE than the corresponding posterior procedure. In addition, increasing fusion lengths were associated with higher risk of VTE. Revision fusions and the use of interbody www.spinejournal.com
915
Spine
SURGERY
Venous Thromboembolism After Spine Surgery • Schairer et al
Postoperative Venous Thromboembolism By Indicating Diagnosis
TABLE 4.
HkV^lÏMFRTl
12.5%10.0%-
VTE Rate
J-' 7.5%-
Hazard Ratio
95% CI
p
Complication
1.45
1.18-1.78
SPINE Volume 39, Number 11, pp 911-918 ©2014, Lippincott Williams &c Wilkins
SURGERY
Venous Thromboembolism After Spine Surgery William W. Schairer, MD,* Andrew C. Pecltke, MD,t and Serena S. Hu, MD+
Study Design. Retrospective cohort study. Objective. To measure the rate of postoperative venous thromboembolic events (VTE) after spine decompression and fusion procedures. Summary of Background Data. VTE after spine surgery is a serious complication, but chemoprophylaxis is not without significant risk due to the concern of epidural hematoma. Current literature report widely variable rates of VTE, and have weaknesses in sample size, specificity of diagnosis, and methodological problems with adequate patient follow-up. Methods. State-level inpatient, ambulatory surgery, and emergency department administrative databases were used to track patients for clinically significant VTE within 90 days of discharge after a spine procedure. Results. Of 357,926 patients enrolled, one-third underwent spine decompression alone, whereas two-thirds received a spine fusion. The overall rate of VTE was 1.37% (95% Cl: 1.33-1.41 ), but varied widely depending on diagnosis, 1.03% for structural degenerative diagnoses to 10.7% for spine infection. Posterior cervical fusion had a higher rate of VTE than anterior cervical fusion, whereas anterior thoracolumbar and lumbosacral fusions had higher rates than the respective posterior approaches. Additional risk factors included patients receiving long spine fusions and having multiple procedures during the hospitalization. Eorty percent of VTEs discovered after discharge were diagnosed at a different hospital. Conclusion. The rate of spine VTE varies widely depending on diagnosis and procedure. It is important to risk-stratify patients who present for spine surgery to identify patients at increased risk who should be monitored for the development of VTE. It is important to know that nearly half of VTEs that occur after discharge are diagnosed at different hospitals, and thus the primary surgeon may From the *Department of Orthopaedic Surgery, Hospital for Special Surgery, New York, NY; tDepartment of Orthopaedic Surgery, University of California San Francisco, San Francisco, CA; and tDepartment of Orthopaedic Surgery, Stanford University School of Medicine, Redwood City, CA. Acknowledgment date: Septemher 16, 2013. First revision date: December 8, 2013. Second revision date: February 1, 2014. Third revision date: February 23, 2014. Acceptance date: February 25, 2014. The manuscript submitted does not contain information about medical device(s)/drug(s). No funds were received in support of this work. Relevant financial activities outside the submitted work: consultancy, payment for lecture, royalties, stocks, travel/accommodations/meeting expenses. Address correspondence and reprint requests to Serena S. FHu, MD, Department of Orthopaedic Surgery, University of California San Francisco, 500 Parnassus Ave, M U 320W, San Francisco, CA 94143-0728; F-mail: HuS@ orthosurg.ucsf.edu DOI: 10.1097/BRS.0000000000000315 Spine
be initially unaware of the complication. These results from a large selection of historical patients may provide a tool for estimating patient risk depending on diagnosis and type of procedure. Key words: spine surgery, decompression, fusion, deep venous thrombosis, venous thromboembolic event, pulmonary embolism, PE, DVT, VTE, complication. Level of Evidence: 2 Spine 2014;39:911-918
V
enous thromboembolism (VTE), which includes deep venous thrombosis (DVT) and pulmonary embolism (PE), are serious complications after surgery. Although in general there are consensus guidelines about early postoperative pharmacological prophylaxis after most surgical procedures, there is less agreement about VTE prophylaxis methods after spine surgery. In addition to an increased risk of infection, anticoagulation leads to increased venous bleeding and increases the risk of a potentially catastrophic neurological complication due to formation of an epidural hematoma compressing the spinal cord.'"^ In addition, aspirin, even when discontinued a week preoperatively, has been shown to cause increased blood drainage and increased transfusion requirements.'' Not surprisingly, there is a wide variability in practice among spine surgeons regarding the use and timing of VTE prophylaxis," as well as wide variation across hospitals.** In the literature, there is a wide range of reported rates of VTE after spine surgery, from as low as 0.29% and up to 23.4%.'*"'^ This disparity is due to a number of factors. One reason is the different rates reported in patients with different indications for spine surgery. In addition, some studies define VTE as only those that are detected clinically, whereas others screened all patients to detect subclinical events. A major methodological flaw is the lack of appropriate follow-up, given that patients seeking urgent and emergent care may not present to the hospital where they underwent surgery. Einally, different time periods for follow-up may significantly impact the number of events detected. Although VTEs are a well-known complication, accurate assessment of the prevalence is of the utmost importance in patients who undergo spine surgery. The American College of Chest Physicians state that patients with risk factors who are undergoing elective spine surgery should receive postoperative low-dose unfractionated heparin or low-molecular weight heparin." However, in their consensus clinical guidelines for antithrombotic therapy in spine surgery. The North American Spine Society was unable to agree on the rate of www.spinejournal.com
911
SURGERY
clinically significant VTE due to the limitations in published data.^ One must weigh the risks and benefits of postoperative chemoprophylaxis, and the risks may be very different depending on the type of procedure and individual patient diagnosis. As such, the North American Spine Society clinical guidelines state that there are not enough high-quality studies with subgroup analyses that allow identifying the risk in different patient groups. Thus, this study set out to address the shortcomings in the current literature regarding the rate of VTF after spine surgery. This study was designed to use a large patient cohort that would allow appropriate subgroup analyses for patients with different diagnoses and procedures. In addition, patient follow-up is the key to accurate diagnoses; this study allowed us to track patients to any hospital within their home state to identify clinically significant VTE up to 6 months after spine surgery. In addition, the secondary purpose of this study was to identify risk factors for VTE depending on patient characteristics and procedure details.
Venous Thromboembolism After Spine Surgery • Schairer et al
Data Sources SID Database
California, Fiorida 2005 - 2009
SASD Database
TT Spine Decompression or Fusion
Group Seiection
387,924 patients Insufficient Follow-up: 21,653 patients
Out-of-state: 8345 patients 357,926 patients
Outcome Measurement
SID Database
/
\ \
SEDD Database
/ 90-day follow-up
Inpatient or ED diagnosis of VTE Figure 1 . Data sources from California and Florida: 2005-2009.
MATERIALS AN D METHODS Data Sources We used the State Inpatient Database (SID), State Ambulatory Surgery Database (SASD), and the State Emergency Department Database (SFDD). These administrative claims databases are maintained by the Healthcare Cost and Utilization Project, which is part of the Agency for Healthcare Research and Quahty. The databases contain 100% of patient records for each available state. Each visit contains information on patient demographics. International Classification of Diseases, Ninth Revision (ICD-9) diagnosis codes, and ICD-9 and Current Procedure Terminology codes (Current Procedure Terminology only in SASD, 7CD-9 in SASD and SID). We compiled the SID, SASD, and SEDD databases from California (2005-2009) and Florida (2005-2009). These states and years were chosen because they participated in all 3 databases (SID, SASD, and SEDD) and they contained a unique identifier to track patient visits between databases and over time. This ensured that we could identify patients with clinically significant VTE whether they returned to the same hospital or went to a different hospital (Figure 1).
Patient Selection Procedure codes were used to identify patients who underwent spine fusion and nonfusion procedures either as an inpatient (in the SID) or outpatient (in the SASD). Patients were excluded if their home state was different than the state of the hospital where they underwent their spine surgery. This was done because in an acute setting, these patients may first seek medical care in their home state and would thus not be captured in our data. In addition, patients with less than 90 days of available follow-up (spine surgery in October, November, or December of 2009) were excluded. Surgical procedures were identified as cervical, thoracolumbar, or lumbosacral. Patients were grouped by 1 of 5 primary diagnoses: structural (spondylosis, spondylolisthesis, 912
www.spinejournal.com
scoliosis, other), complication of implant or previous surgery, trauma (including vertebral compression fractures), infection, and cancer. Medical risk factors associated with VTE were identified using ICD-9 diagnosis codes, including some groups from the Elixhauser medical comorbidity definitions.''' These risk factors were included in a multivariate model to increase the specificity of risk associated with each diagnosis and procedure.
Identification of Venous Thromboembolic Events Clinically relevant DVT and PE were identified by ICD-9 diagnosis codes. Events occurred during the index hospital visit (SID or SASD databases) or within 90 days after discharge (SID or SEDD databases). To isolate VTE risk related to spine surgery, patients who were readmitted to the hospital or presented to the emergency department were censored from that point forward. In addition, patients were censored after 1 VTE event to avoid counting patients more than once.
Statistical Analysis
The incidence of VTE occurring within 90 days of discharge was evaluated using a Kaplan-Meier time-to-event analysis; groups were compared with a log-rank test. Event rates are presented both as the Kaplan-Meier estimation and as the absolute number of events. Risk factors were evaluated using a Cox proportional hazards model, and included patient demographics, surgical procedure, and diagnosis. To better evaluate the different spine fusion procedures, each was compared with decompression alone. In addition, multivariate models were stratified by medical comorbidities related to adjust for different baseline risk for each patient (history of DVT, atrial fibrillation, use of anticoagulation medication, coagulation deficiency, current tobacco use, peripheral vascular disease, pregnancy, obesity, coronary artery disease, lymphoma, and cancer). May 2014
Spine
SURGERY
Venous Thromboembolism After Spine Surgery • Schairer et al
C O
U
^
LO
ro
O
o
ro O
O
rM
O
LO
^
I—
LO
ro O
CN
CO
ro
CN
LO
CO
CO
LO
tn
IX
tn
o
tN
LO
CO
tN
O
o
O
evic Sur mpl Itio
s? ai .w
o Q c
LO
CO
I—
O
ro
tn O
CO
o
O CN
tN
r-
tN
o "^
CN
LO
ro
O
ro
CO
tN
IX
O
O
ro
D Uo 01
IX
E 5^ o
tn
CO CO
tN
ro
CO
CO
O
U
¡^
•
V.D
*"
3 ^^
ro
LO
t í S?
,—
tn
LO CTl
O
O
LO
c
O
en
CO
CO tN
Ix .^ 0^
CO CN
^
ro O
ro
tN
ro
LO
LO
rN
IX
tN
LO
m
^ tN
Ix ,
Ix
tN
tN
tN
CO tN
ro
ro ro
CN
tN
IX ,
ro
ro
LO
CTi CO
LO
O^
Oi
O
LO
CM
ro
LO
o '-D IX
tN
d
LO
LO
o^
^
Ix
LO
LO
tji
O CN
IX
CO
O
ro
CN ,—
en
o
CN IX
O
o
CO
t-M
ix'
O
LO
rM
O
,
CN
tN
tN
o
O
O
O
O
LO
ro
o
tJi
LO
LO
>—
r—
tN
LO
ro
tN
ro ro
rM
CO LO
ro
CN
tN
(Jl
tN tN
en
00 IX
u O
o
5, lumbc
•s ;""
t3^
4.6
ro
IX -—
r—
2.4
os
r—
MuH Stage
L
^
16.3
o
12.3
S 's
8.3
ity
3
IX
ro
O
5 Ix
Ix
ro
CN
IX
CTi
en
ro
•~O
LO
ro
CN
LO
O
°^ E 5^
ro O
ixi
tn
tTi
LO
ro
CN LO
IX tN LO
^
CO
ro "*
, LO
CO ^
LO
CN
O
CO
CN LO
tTi
LO
.^
"*
CO
tn
CO
IX
CN LO
LO LO
tN LO
LO
fN
CN CO Ü1
LO
tn
tN
ro CO
iri
v£)
tTi
tji LO
CN LO
CO CM
K tN
ro
CN"
ro ro
ro O tn
ro ro LO
CN
nterio r-posti rior
ro
LO
(Ti
a."
LS posterior i usion 1
o
LS AP fusion
LS anter 1
sion
LS déco Tipre ssion 1
rz
Lumbosac
TL posterior 1 usion 1
TL AP fusion
TL anter 1
sion
TL décompre ssion
o
1
Cervica
Q.
n]ïïK-. 1ndicating Diagnosis Structural
Trauma
Cancer
Device/Surgical Complication
Infection
Cervical decompression
0.8 (0.7-1.0)
4.8(1.8-12.2)
8.9(5.3-14.6)
8.4(4.8-14.4)
4.5(1.7-11.7)
Atlas-axis fusion
1.8(1.1-2.9)
1.5(0.2-10.3)
5.5(4.1-7.4)
No events
2.9(0.4-19.1)
Cervical anterior fusion
0.4 (0.3-0.4)
0.7 (0.3-1.6)
3.9(3.3-4.7)
9.8(6.6-14.4)
5.5(3.0-10.2)
Cervical posterior fusion
1.5 (1.3-1.9)
0.6 (0.2-1.8)
5.9(4.8-7.1)
7.7(3.0-19.2]
5.2 (3.1-8.8)
Cervical AP fusion
2.8(2.1-3.6)
2.7 (0.9-8.2)
9.5(7.7-11.7)
6.4(2.3-17.1)
4.7(1.8-11.9)
TL decompression
1.0(0.8-1.3)
1.7(0.4-6.5)
2.3(1.0-5.6)
11.9(7.7-18.3)
2.6 (1.0-6.7)
TL anterior fusion
2.1 (1.2-3.6)
4.8 (0.7-29.3)
6.1 (4.0-9.1)
9.2 (4.7-17.5)
7.3 (4.1-13.1)
TL posterior fusion
0.9(0.7-1.2)
2.4(1.3-4.4)
7.0(6.1-8.1)
7.5(4.3-13.0)
6.5 (4.9-8.6)
TL AP fusion
3.3(2.1-5.1)
5.4(1.4-19.9)
8.2(5.2-12.7)
14.6(9.0-23.0)
10.1 (5.7-17.5)
LS decompression
0.6 (0.5-0.6)
2.3(1.3^.2)
3.1 (2.1^.4)
6.9 (5.0-9.5)
6.1 (3.5-10.6)
LS anterior fusion
1.5 (1.3-1.8)
0.9 (0.3-2.8)
3.6(2.0-6.3)
5.6(2.2-14.3)
6.8(2.5-17.4)
LS posterior fusion
1.1 (1.0-1.1)
1.4 (1.0-1.9)
3.3 (2.6-4.2)
9.1 (5.6-14.5)
5.9(3.8-9.1)
LSAP fusion
1.6(1.4-1.9)
3.6(2.3-5.8)
4.8(2.7-8.5)
11.1 (6.0-20.0)
10.9(5.3-21.8)
2.9 (2.3-3.8)
3.1 (1.4-6.8)
10.1 (7.6-13.3)
11.5(6.2-21.0)
12.7(8.2-19.2)
Procedure
.„^•_xi.r..,
Cervical
Thoracolumbar
Lumbosacral
Cross junctional (cerv/TL, TL/LS)
The vaiues given are Kapian-Meier % (95% confidence intervais). AP indicates anterior-posterior; TL, thoracolumbar; LS, Iumbosacrai; CI, confidence intervai.
RESULTS Demographics We identified 357,926 patients who underwent a spine decompression and/or fusion (Table 1). Thirty-one percent of the procedures were cervical, 6% were thoracolumbar, and 62% were lumbosacral; 1 % of the patients had fusions extending across a junctional level (cervical and thoracolumbar fusion or thoracolumbar and lumbosacral fusion). Decompression procedures made up one-third of procedures, whereas twothirds of procedures included spine fusion. Overall, 2.7% of spine fusions were revision procedures. Structural diagnoses accounted for the vast majority of diagnoses. Trauma accounted for a significant portion of cervical and thoracolumbar diagnoses, and infection and cancer were a prominent diagnosis in the thoracolumbar group. Venous Thromboembolic Events The overall rate of VTE was 1.37% (Kaplan-Meier 95% CI: 1.33-1.41; absolute N = 4657/357,926). DVT occurred in 1.07% of cases (1.04-1.11) (Table 2), whereas PE occurred in 0.45% (0.43-0.47) of cases (Table 3). VTE occurred in 0.92% (0.87-0.98) of decompression cases, 1.10% (1.04-1.17) of cervical fusion cases, 4.54% (4.20-4.91) of thoracolumbar 914
www.spinejournal.com
fusion cases, and 1.74% (1.66-1.82) of lumbosacral fusion cases (Eigure 2). Patients who received longer spine fusion had higher rates of VTE. In short fusions (2-3 vertebrae), 1.29% (1.23-1.34) of patients had a VTE (hazards ratio [HR] = 1.41, ? < 0.001, compared with decompression). Medium (4—8 vertebrae) and long (9 or more vertebrae) spine fusions had VTE rates of 3.05% (2.88-3.23) (HR = 3.45, ? < 0.001) and 2.80% (2.38-3.28) (HR = 3.22, P < 0.001), respectively. Procedures where the patient was in the supine position were associated with a VTE rate of 0.87% (0.81-0.93). However, patients that were in the prone position had a higher rate of 1.55% (1.50-1.60) (HR = 1.77, P < 0.001). There was wide variation in VTE rate depending on diagnosis (Eigure 3). At the time of discharge from the initial hospitalization, the VTE rate of the structural and implant/surgical complication groups was 0.45% (0.42-0.47) and 0.94% (0.74-1.20), respectively. The trauma, infection, and cancer groups were higher at 4.12% (3.81-4.46), 7.72% (6.698.92), and 4.56% (3.88-5.36). At 90 days after discharge, the overall VTE rates were 1.02% (0.99-1.06) and 1.94% (1.63-2.32) for structural and implant/surgical complication, respectively. As with at discharge, the trauma, infection, and cancer groups were higher at 90 days with VTE rates May 2014
SURCERY
"TOLE
Venous Thromboembolism After Spine Surgery • Schairer et al
s^iiisraigryjggjgi^ïTîiJîiirnïïmTirasai^ Indicating Diagnosis Structural
Trauma
Cancer
Device/Surgical Complication
Infection
Cervical decompression
0.4 (0.3-0.6)
1.1 (0.2-7.6)
2.7(1.1-6.3)
5.4(2.6-11.2)
4.9(1.8-12.7)
Atlas-axis fusion
0.5(0.2-1.3)
1.5 (0.2-10.1)
0.9 (0.4-1.9)
No events
2.9(0.4-19.1)
Cervical anterior fusion
0.2 (0.1-0.2)
0.5(0.2-1.2)
1.4(1.0-1.9)
2.3(1.1-5.1)
2.7(1.1-6.6)
Cervical posterior fusion
0.7 (0.5-0.9)
0.2 (0.0-1.4)
1.5(1.0-2.3)
1.9(0.3-12.9)
2.0 (0.8-4.8)
Cervical AP fusion
0.7(0.4-1.2)
1.9(0.5-7.5)
1.5(0.9-2.6)
2.1 (0.3-14.2)
No events
TL decompression
0.5 (0.3-0.7)
No events
2.1 (0.9-5.1)
5.2(2.5-10.6)
No events
TL anterior fusion
1.3(0.6-2.8)
4.8 (0.7-29.3)
1.8(0.8-3.7)
2.3 (0.6-8.8)
1.3 (0.3-5.3)
TL posterior fusion
0.3 (0.2-0.5)
0.5(0.1-2.1)
2.3(1.7-3.0)
4.6(2.3-9.1)
2.7(1.8-4.1)
TLAP fusion
1.5(0.8-3.0)
No events
3.2(1.6-6.4)
2.4 (0.8-7.4)
4.2(1.7-10.2)
LS decompression
0.3 (0.2-0.3)
0.4(0.1-1.7)
0.7(0.3-1.5)
3.1 (1.9-5.0)
3.3(1.5-7.4)
LS anterior fusion
0.5 (0.3-0.7)
No events
2.3(1.2-4.6)
1.4(0.2-9.6)
4.6(1.1-18.4)
LS posterior fusion
0.6 (0.5-0.6)
0.6 (0.4-1.0)
1.0(0.7-1.6)
No events
1.6(0.7-3.7)
LSAP fusion
0.7 (0.6-0.9)
0.7(0.2-2.1)
2.5(1.2-5.6)
No events
7.5(2.8-19.3)
1.5(1.0-2.1)
No events
3.5 (2.1-5.7)
4.7(1.5-14.5)
7.3(3.7-14.1)
Procedure Cervical
Thoracolumbar
Lumbosacral
Cross junctional (cerv/TL, TL/LS)
The values given are Kaplan-Meier pereent (95% eonfidenee intervais). AP indicates anterior-posterior; TL, thoraeolumbar; LS, lumbosacral; eerv, cervical.
of 6.41% (6.01-6.85), 10.67% (9.37-12.15), and 8.26% (7.24-9.41). Timing of Events Overall, 52.7% (n = 2454) of VTEs occurred during the index hospitalization, whereas 47.3% (n = 2203) occurred after
Postoperative Venous Thromboembolism
discharge (median time, 9.1 d; standard deviation 19.6 d). Events diagnosed during the index hospitalization were PEs in 29.6% of cases compared with 35.9% of cases diagnosed after discharge (P < 0.001). The mortality rate of PE diagnosed during the initial hospitalization was 4.0%, whereas it was lower if diagnosed after discharge at 2.6%. In addition, 40% of VTEs found after discharge were diagnosed at a different hospital than where the initial procedure was performed.
By Procedure Type
Multivariate Model
5.0%-I
4.0%-
3,0% H
^ ^ ----• ^— ^ " -
2.0%-
Thoracolumbar fusion Lumbosacral fusion Cervical fusion Decompression
1.0%-
0.030 60 Days since discharge
90
Figure 2. Postoperative venous thromhoembolism by procedure type. Spine
Independent risk factors assessed through a multivariate model that controlled or a number of medical comorbidities that may affect VTE risk. The model showed an increasing risk of VTE with increasing age (Table 4). In addition, male sex was associated with an increased risk compared with female sex. For the type of surgical procedure, compared with decompression alone, all fusion procedures had higher risk of VTE with the exception of atlas-axis fusion. Although anterior cervical fusion had a lower risk than posterior cervical fusion, the anterior thoracolumbar and lumbosacral fusions had greater risk of VTE than the corresponding posterior procedure. In addition, increasing fusion lengths were associated with higher risk of VTE. Revision fusions and the use of interbody www.spinejournal.com
915
Spine
SURGERY
Venous Thromboembolism After Spine Surgery • Schairer et al
Postoperative Venous Thromboembolism By Indicating Diagnosis
TABLE 4.
HkV^lÏMFRTl
12.5%10.0%-
VTE Rate
J-' 7.5%-
Hazard Ratio
95% CI
p
Complication
1.45
1.18-1.78
