Do Prophylactic Inferior Vena Cava Filters Affect the Rates of Venous Thromboembolism and Pulmonary Embolism in Patients Undergoing Major Spine Surgery?

Abstract

Study Design

Retrospective Analysis

Background

Venous thromboembolism (VTE) represents a significant cause of morbidity and mortality in major spine surgery. Placement of prophylactic inferior vena cava filters (IVCF) in patients undergoing major spine surgery was previously adopted at our institution. This study reports our experience and compares VTE rates between patients with and without preoperative IVCF placement.

Methods

A Retrospective comparative study was conducted on adult patients who underwent IVCF placement and those who did not prior to their spinal fusion procedure, between 2013 and 2016. Thoracolumbar fusions (anterior and/or posterior) of 7 or more levels, spinal osteotomies, and a minimum of a 3-month follow-up were included. Traumatic, oncologic, and cervical pathology were excluded. Primary outcomes measured included the incidence of overall VTE (DVT/PE), death, IVCF related complications, and IVCF retrieval.

Results

386 patients who underwent major spine surgery, 258 met the eligibility criteria. Of those patients, 105 patients (40.7%) had prophylactic IVCF placement. All patients had postoperative SCDs and chemoprophylaxis. The presence of an IVCF was associated with an increased rate of overall VTE (14.3% vs 6.5%, P ≤ .05) and DVT episodes (8.6% vs 2.6%, P = .04). The rate of PE for the IVCF group and non-IVCF group was 8.6% and 4.6%, respectively, which was not statistically significant (P = .32). The all-cause mortality rate overall of 2.3% was statistically similar between both groups (P = 1.0). The IVCF group had higher rates of hematoma/seroma vs the non-IVCF group (12.4% vs 3.9%, P ≤ .05). 99 IVCFs were retrievable designs, and 85% were successfully retrieved. Overall IVCF-related complication rate was 11%.

Conclusions

No statistical difference in PE or mortality rates existed between the IVCF and the control group. Patients with IVCF placement experienced approximately twice the rate of VTE and three times the rate of DVT compared to those without IVCF. The IVCF-related complication rate was 11%. Based on the results of this study, the authors recommend against the routine use of prophylactic IVCFs in adults undergoing major spine surgery.

Level of Evidence

III.

Introduction

Venous thromboembolism (VTE) events, including deep venous thrombosis (DVT) and pulmonary embolism (PE), are well-documented causes of morbidity and mortality in orthopaedic surgery.¹ Increased VTE risk often is attributed to decreased mobilization postoperatively and in patients undergoing spine surgery, that is, due to direct manipulation of the large retroperitoneal veins in anterior approaches.^2-5 A systematic review of the spine literature estimates the risk of postoperative DVT at approximately 2.1% and PE at .3%, but reported rates from individual studies vary widely.⁶ A recent prospective study by the International Spine Study Group examined adult patients undergoing major deformity surgery and reported DVT and PE rates of 2.1% for both.⁷

When considering VTE chemoprophylaxis, the surgeon must weigh risks of symptomatic VTE against the risk of neurologic injury from an epidural hematoma. Cain et al⁸ found significant morbidity associated with therapeutic anticoagulation postoperatively. A survey of practicing spine surgeons found that epidural hematoma is perceived to be a substantial risk with chemoprophylaxis.⁹ Given these findings, multiple studies have addressed placement of prophylactic inferior vena cava filters (IVCF) to prevent symptomatic PE.^2,10-13 These studies generally support the efficacy and safety of prophylactic IVCF placement. Still, they are notably retrospective in nature, lack adequate controls, and are potentially underpowered. Further, complications directly associated with IVCF placement/retrieval are more common than previously thought and remain an area of active investigation.^14-17

At our institution, prophylactic IVCF filter placement was implemented in a significant portion of patients undergoing major spine surgery. The aim of this retrospective study was 3-fold: (1) report rates of symptomatic postoperative VTE (DVT/PE) and mortality, (2) compare postoperative symptomatic VTE rate between patients with and without prophylactic IVCF, and (3) report IVCF related complications and retrieval rate.

Methods

After Institutional Review Board approval, the University of Pennsylvania Health System’s electronic medical record (EMR) (Epic System Corp., Verona, WI) was queried by CPT code for all patients who underwent “major spine surgery” between January 2013 and August 2016. Major spine surgery procedures included: anterior spine fusion (CPT codes 22558, 22585), instrumented posterior spinal fusion of seven or more contiguous levels (CPT codes 22802, 22804, 22843, 22844), or osteotomy for deformity correction (CPT codes 22212, 22212, 22206, 22207, 22208).

The electronic record of each patient was queried for associated VTE diagnoses after the date of surgery using ICD 9/10 codes for VTE, pulmonary embolism, and cardiac arrest. Inpatient orders were queried for tests associated with VTE diagnosis, including CT chest with contrast, duplex ultrasound of the lower extremity, ventilation-perfusion scan, and venogram. Inpatient orders were searched for initiation of VTE prophylaxis treatment modalities, including; sequential compression devices, TED stockings, unfractionated heparin, low molecular weight heparin, warfarin, aspirin, clopidogrel, ticagrelor, and rivaroxaban. Each subject’s record was searched for a CPT code associated with IVCF placement (CPT codes 37191, 37192, 37193).

Each chart was manually reviewed by independent observers to determine study eligibility and to extract data. Inclusion criteria included: age ≥18 years old at the time of surgery, preoperative medical/cardiology evaluation, intraoperative or immediate postoperative radiograph confirming the presence/absence of an IVCF, postoperative admission to the orthopaedic or neurosurgical service (2-day minimum), and a minimum of 3 months clinical follow-up with the attending surgeon. Exclusion criteria included: cervical surgery, acute spinal trauma, active spine infection, or spinal tumor. Eligible patients’ charts were systematically reviewed for demographic and clinical variables of interest, postoperative diagnosis of VTE, presence of IVCF on a perioperative radiograph, and IVCF-associated complications. Charts were examined for VTE risk factors, including history of VTE, cancer, or diagnosis of thrombophilia. Obesity and anterior spine surgery were not included as VTE risk factors but were collected independently. If the 3-month follow-up minimum was not met, the Social Security Death Index was queried to determine if they had died.

Demographic and clinical variables were reported with descriptive statistics. Univariate statistical analyses were conducted using Pearson’s chi-square test or Fisher’s exact test for categorical or binary variables, and Student’s t-test for continuous variables. A cutoff for significance, P ≤ .10, was used for entry into the multiple regression model. Using statistically significant variables from the univariate analysis and a backward likelihood ratio technique, a multivariate logistic regression model was formulated to determine the variables that were independent risk factors for VTE. Simpler models were generated from the original with continuous variables and compared against the more robust models using the −2 log-likelihood method. When the final model was derived, we performed an area under the curve test to assess the predictive power of the model. Statistics were calculated using SPSS (v 22.0 - SPSS Inc., Chicago, IL).

Results

The medical record query yielded 384 unique patients. 258 patients were eligible, and 126 were excluded: 77 for cervicothoracic surgery, 16 for acute spinal trauma, 6 for active infection, 7 for spinal tumor, 13 cases were deemed not “major spine surgery,” and 7 cases for inadequate follow-up. No patients lost to follow up died according to the Social Security Death Index. There was no missing data for all 258 included patients.

The remaining 258 patients had no missing data. Characteristics are summarized in Table 1. The orthopaedic spine service performed 81.4% of surgeries. 90.3% of patients had a diagnosis of degenerative etiology, including degenerative scoliosis, stenosis, spondylolisthesis, degenerative disc disease, or failure of prior surgery (“failed back”). Anterior spinal fusion was performed in 68.5% of patients, while anterior and posterior spinal fusion was performed in 55.8%. Surgical variables, including number of levels fused, duration of surgery, and estimated blood loss (EBL) varied widely between patients. One case reported 13 000 mL EBL due to iliac vein injury during the anterior approach. All patients had sequential compressive devices applied at the time of surgery and were continued for the duration of hospitalization. 95% received subcutaneous heparin or enoxaparin VTE chemoprophylaxis within 24 hours after surgery. 21% of patients had an identifiable preoperative risk factor for VTE. The rate of hematoma/seroma was 7.4%, and the rate of unplanned return to the operating room during the follow-up period was 21.3%. Twenty-five patients had symptomatic, diagnostic-confirmed VTE, resulting in a 9.9% overall VTE rate. The DVT and PE rates were 5% and 5.8%, respectively. The overall mortality rate during the follow-up period was 2.3%.

Table 1.

Overall patient characteristics.

Variable
Total subjects, n	258
Female, n (%)	160 (62.0%)
Non-white race, n (%)	49 (19.0%)
Age at time of surgery, mean (range), years	54.5 (18–86)
Age at time of surgery, n (%), years
<65 years	183 (70.9%)
65+ years	75 (29.1%)
Body mass index, mean (range)	27.4 (17.9–46.8)
Body mass index, n (%)
<30 BMI	179 (69.4%)
30+ BMI	79 (30.6%)
American Society anesthesia score, n (%)
ASA 1 or 2	184 (71.3%)
ASA 3+	74 (28.7%)
Hospital service, n (%)
Neurosurgery	48 (18.6%)
Orthopaedic surgery	210 (81.4%)
Pathology, n (%)
Deformity, primary	79 (30.6%)
Degenerative (stenosis, deformity, DDD)	233 (90.3%)
Approach for spinal fusion, n (%)
Anterior spine fusion	177 (68.6%)
Combined anterior-posterior fusion	144 (55.8%)
Posterior spine fusion only	81 (31.4%)
Staged surgery, n (%)	57 (22.1%)
Levels fused, mean (range)	7.1 (2–18)
Levels fused, n (%)
7+ levels	131 (50.8%)
13+ levels	38 (14.7%)
Iliac fixation, n (%)	99 (38.2%)
Osteotomy performed, n (%)	103 (39.9%)
Revision surgery, n (%)	108 (41.9%)
Surgery duration, mean (range), hours	5.75 (1.0–11.6)
Surgery duration, n (%)
<6 hours	143 (55.4%)
>6 hours	115 (44.6%)
Tranexamic acid use, intraoperative, n (%)	59 (22.7%)
Estimated blood loss, mean (range), mL	1146 (10–13 000)
Estimated blood loss, n (%)
<1000 mL	150 (58.1%)
>1000 mL	108 (41.8%)
Chemoprophylaxis on POD 0 or 1, n (%)	245 (95.0%)
Unplanned return to OR during f/u period, n (%)	55 (21.3%)
Length of stay, n (mean), day	6.8 (2–44)
Length of stay, n (%)
<7 days	164 (63.6%)
7+ days	94 (36.4%)
Disposition, n (%)
Home with healthcare services	160 (62.0%)
Subacute nursing facility	44 (17.1%)
Inpatient rehabilitation or LTACH	52 (20.2%)
Inpatient death	2 (.8%)
VTE risk factor identified in history, n (%)	34 (21.5%)
Prophylactic IVCF use, n (%)	105 (40.7%)
Postoperative hematoma/seroma, n (%)	19 (7.4%)
Venous thromboembolism, n (%), subjects	25 (9.9%)
Specific venous thromboembolism, n (%), events
Deep venous thrombosis	13 (5.0%)
Pulmonary embolism	15 (5.8)
Death during f/u period, n (%)	6 (2.3%)

The patients were stratified based on prophylactic IVCF placement preoperatively, and characteristics of both cohorts are summarized in Table 2 with associated univariate hypothesis testing for each variable. 105 eligible patients (40.7%) had an IVCF placed preoperatively. Patients with higher BMI, anterior spinal fusion, staged procedures, revision, and advanced age were more likely to receive an IVCF preoperatively. No significant difference existed in terms of age, ASA classification, number of levels fused, osteotomy performed, surgery duration, EBL, or unplanned return to the operating room. Notably, tranexamic acid (TXA) use and initiation of chemoprophylaxis within 24 hours postoperatively showed no significant differences. Patients receiving an IVCF were more likely to have a VTE risk factor. Postoperative hematoma/seroma was more common amongst patients with an IVCF. In this univariate analysis, the presence of an IVCF was associated with an increased rate of overall VTE (14.3% vs 6.5%, P ≤ .05) and DVT episodes (8.6% vs 2.6%, P = .04). There was no difference in PE rate or mortality rate between the two groups.

Table 2.

Patients Stratified by Presence of a Prophylactic IVCF.

	No Prophylactic	Prophylactic	P value
Variable	IVCF	IVCF	P value
Total subjects, n	153	105
Female, n (%)	94 (61.4%)	66 (62.9%)	.88
Non-white race, n (%)	28 (18.3%)	21 (20%)	.76
Age at time of surgery, mean (range), years	52.1 (18–86)	58.0 (28–77)	<.01*
Age at time of surgery, n (%), years
<65 years	106 (69.3%)	77 (73.3%)
65+ years	47 (30.7%)	28 (26.7%)	0.7
Body mass index, mean (range)	26.8 (17.9–46.8)	28.3 (18.6–39.9)	.03*
Body mass index, n (%)
<30 BMI	116 (75.8%)	63 (60%)
30+ BMI	37 (24.2%)	42 (40%)	.02*
American Society anesthesia score, n (%)
ASA 1 or 2	110 (71.9%)	74 (70.5%)
ASA 3+	43 (28.1%)	31 (29.5%)	.83
Hospital service, n (%)
Neurosurgery	43 (28.1%)	5 (4.8%)
Orthopaedic surgery	110 (71.9%)	100 (95.2%)	.04*
Pathology, n (%)
Deformity, primary	56 (36.6%)	23 (21.9%)	.04*
Degenerative (stenosis, deformity, DDD)	131 (85.6%)	102 (97.1%)	.34
Approach for spinal fusion, n (%)
Anterior spine fusion	80 (52.3%)	97 (92.4%)	<.01*
Combined anterior-posterior fusion	53 (34.6%)	91 (86.7%)	<.01*
Posterior spine fusion only	73 (47.7%)	8 (7.6%)	<.01*
Staged surgery, n (%)	16 (10.5%)	41 (39%)	<.01*
Levels fused, mean (range)	7.1 (2–18)	7.2 (2–17)	.89
Levels fused, n (%)
7+ levels	85 (55.6%)	46 (43.8%)	.19
13+ levels	21 (13.7%)	17 (16.2%)	.61
Iliac fixation, n (%)	47 (30.7%)	52 (49.5%)	.02*
Osteotomy performed, n (%)	60 (39.2%)	43 (41%)	.83
Revision surgery, n (%)	51 (33.3%)	57 (54.3%)	.01*
Surgery duration, mean (range), hours	5.6 (1.0–11.5)	6.0 (1.9–11.6)	.12
Surgery duration, n (%)
<6 hours	86 (56.2%)	57 (54.3%)
>6 hours	67 (43.8%)	48 (45.7%)	.82
Tranexamic acid use, intraoperative, n (%)	35 (22.9%)	24 (22.9%)	.99
Estimated blood loss, mean (range), mL	1154 (10–6000)	1135 (100–13 000)	.91
Estimated blood loss, n (%)
<1000 mL	92 (60.1%)	58 (55.2%)
>1000 mL	61 (39.9%)	47 (44.8%)	.55
Chemoprophylaxis on POD 0 or 1, n (%)	143 (93.5%)	102 (97.1%)	.77
Unplanned return to OR during f/u period, n (%)	29 (19%)	26 (24.8%)	.32
Length of stay, n (mean), day	6.2 (2–43)	7.7 (3–44)	.04
Length of stay, n (%)
<7 days	108 (70.6%)	56 (53.3%)
7+ days	45 (29.4%)	49 (46.7%)	.02
VTE risk factor identified in history, n (%)	13 (8.5%)	21 (20%)	.01
Postoperative hematoma/seroma, n (%)	6 (5.7%)	13 (12.4%)	.01
Venous thromboembolism, n (%), subjects	10 (6.5%)	15 (14.3%)	<.05
Specific venous thromboembolism, n (%), events
Deep venous thrombosis	4 (2.6%)	9 (8.6%)	.04 (Fisher’s exact)
Pulmonary embolism	7 (4.6%)	8 (7.6%)	.32
Death during f/u period, n (%)	4 (2.6%)	2 (1.9%)	1.00 (Fisher’s exact)

Next, patients were stratified based on symptomatic postoperative VTE, and the characteristics of these two cohorts are summarized in Table 3 with associated univariate hypothesis testing for each variable. Stratification by VTE was done because individual DVT and PE events were so rare. Twenty-five (9.9%) eligible patients had symptomatic postoperative VTE. Patients with a VTE were older, had more levels fused, longer surgeries, and greater EBL. There were no associations found with BMI, anterior spinal fusion, TXA use, initiating chemoprophylaxis within 24 hours of surgery, or postoperative hematoma/seroma. No patients with a non-degenerative etiology experienced a VTE. Patients with VTE risk factor(s) were more likely to have symptomatic VTE (28% vs 11.6%, P = .03). There was a positive association between prophylactic IVCF use and VTE (60% vs 38.6%, P = .04), and longer duration of IVCF placement (>60 days) and VTE (56% vs 17.2%, P ≤.01).

Table 3.

Patients Stratified by VTE Diagnosis.

Variable	No VTE	VTE	P value
Total subjects, n	233	25
Female, n (%)	148 (63.5%)	12 (48%)	.12
Non-white race, n (%)	45 (19.3%)	4 (16%)	.99
Age at time of surgery, mean (range), years	53.8 (18–86)	61.2 (28–77)	.01*
Age at time of surgery, n (%), years
<65 years	171 (73.4%)	12 (48%)
65+ years	62 (26.6%)	13 (52%)	<.01*
Body mass index, mean (range)	27.3 (17.9–46.8)	28.3 (19.4–40.6)	.38
Body mass index, n (%)
<30 BMI	165 (70.8%)	14 (56%)
30+ BMI	68 (29.2%)	11 (44%)	.13
American Society anesthesia score, n (%)
ASA 1 or 2	169 (72.5%)	15 (60%)
ASA 3+	64 (27.5%)	10 (40%)	.19
Hospital service, n (%)
Neurosurgery	42	6
Orthopaedic surgery	191	19	.75
Pathology, n (%)
Deformity, primary	72 (30.9%)	7 (28%)	.77
Degenerative (stenosis, deformity, DDD)	208 (89.3%)	25 (100%)	.15
Approach for spinal fusion, n (%)
Anterior spine fusion	162 (69.5%)	15 (60%)	.33
Combined anterior-posterior fusion	130 (55.8%)	14 (56%)	.18
Posterior spine fusion only	71 (30.4%)	10 (40%)	.42
Staged surgery, n (%)	49 (21%)	8 (32%)	.21
Levels fused, mean (range)	6.9 (2–18)	9.3 (2–16)	.01*
Levels fused, n (%)
7+ levels	114 (48.9%)	17 (68%)	.07
13+ levels	31 (13.3%)	7 (28%)	<.05*
Iliac fixation, n (%)	85 (36.5%)	14 (56%)	.06
Osteotomy performed, n (%)	90 (38.6%)	13 (52%)	.19
Revision surgery, n (%)	94 (40.3%)	14 (56%)	.13
Surgery duration, mean (range), hours	5.6 (1.0–11.6)	6.9 (4.1–11.5)	<.01*
Surgery duration, n (%)
<6 hours	135 (57.9%)	8 (32%)
>6 hours	98 (41.1%)	17 (68%)	.01*
Tranexamic acid use, intraoperative, n (%)	53 (22.7%)	6 (24%)	NS
Estimated blood loss, mean (range), mL	1038 (10–6000)	2152 (200–13 000)	<.05*
Estimated blood loss, n (%)
<1000 mL	142 (60.9%)	8 (32%)
>1000 mL	91 (39.1%)	17 (68%)	<.01*
Chemoprophylaxis on POD 0 or 1, n (%)	222 (95.3%)	23 (92%)	.37 (Fisher’s exact test)
VTE risk factor identified in history, n (%)	27 (11.6%)	7 (28%)	.03*
Postoperative hematoma/seroma, n (%)	15 (6.4%)	4 (16%)	.10 (Fisher’s exact test)
Prophylactic IVCF use, n (%)	90 (38.6%)	15 (60%)	.04*
IVCF In place >60 days, n (%)	40 (17.2%)	14 (56%)	<.01*

A multivariate regression model was formulated to evaluate independent risk factors associated with symptomatic VTE. Presence of a prophylactic IVCF and other statistically significant variables in the univariate analysis (P ≤ .10) were included, and then a backwards likelihood ratio technique was performed to determine the final model. Statistically significant variables included age >65 (OR 2.90, 95% CI 1.17–7.16, P = .02) and longer duration of IVCF placement (OR 14.73, 95% CI 1.81–119.7, P = .01) Table 4. EBL >1000 mL trended towards significance (OR 2.39, 95% CI .94–6.06, P = .07). In the multiple regression model, the use of a prophylactic IVCF was neither a significant risk factor nor protective for postoperative VTE (OR .33, 95% CI .04–2.71, P = .30).

Table 4.

Multivariate regression model to predict VTE outcome.

Variable	OR	95% CI	P value
Age at time of surgery 65+	2.90	(1.17–7.16)	.02
Estimated blood loss >1000 mL	2.39	(.94–6.06)	.07
Prophylactic IVCF use	.33	(.04–2.71)	0.3
IVCF in place >60 days	14.73	(1.81–119.7)	.01

R square = .105. Area under receiver operating characteristic curve = .820.

Finally, Table 5 displays findings related to IVCF use, including complications. 105 prophylactic IVCFs were placed; 99 (94.3%) were retrievable designs. Of retrievable ICVFs, 84 (84.8%) were successfully retrieved at the most recent follow-up. For the 84 retrieved IVCFs, the average indwelling duration was 78 days (range 21-385). The overall IVCF complication rate was 10.5%: vena cava wall perforation (3 cases), angulation/tilt that complicated placement or retrieval (3 cases), new-onset lymphedema (2 cases), IVCF/vena cava thrombosis (2 cases), cardiac tamponade (1 case), and access vein obstruction (1 case).

Table 5.

Prophylactic IVCF characteristics.

Total prophylactic IVCFs, n	105
IVCF type
Retrievable, n (%)	99 (94.3%)
Permanent, n (%)	6 (5.7%)
Retrievable IVCFs
Retrieved successfully, n (%)	84 (84.8%)
Duration in place, mean (range), days	78 (21-385)
Complications, all IVCFs, n (%)	11 (10.5%)
IVCF complications, by patient
IVCF perforation into vena cava wall, required advanced techniques for removal	1
IVCF perforation into vena cava wall, deemed non-retrievable, developed late DVT	2
IVCF tilt/angulation and perforation into vena cava wall, delayed retrieval required advanced techniques for removal	3
IVCF tilt/angulation, required advanced techniques for removal	4
IVCF tilt/angulation during insertion, required removal and angioplasty for proper orientation	5
IVCF thrombosis noted on cavogram at 1st removal attempt, delayed removal	6
IVCF thrombosis noted on cavogram at 1st removal attempt, no retrieved	7
New-onset lymphedema during the postoperative period	8
New-onset lymphedema during the postoperative period	9
Cardiac tamponade immediately after IVCF removal, required pericardiocentesis and drain placement	10
Access vein (right internal jugular) found to be obstructed, prolonged retrieval through left jugular vein	11

Discussion

Postoperative VTE and associated symptomatic PE represents a serious acute complication of spine surgery that persists for at least 4 weeks postoperatively.¹⁸ Use of any chemoprophylaxis, a concern for postoperative hematoma, specifically a potentially catastrophic epidural hematoma, exists.^8,19-24 The perceived risk⁹ of anticoagulant-associated hematoma has prompted prophylactic IVCF placement. The most common accepted indication for placement is a patient with an absolute contraindication for anticoagulation.²⁵ Other indications, such as VTE prophylaxis in high-risk patients, have expanded in recent years but lack strong supporting evidence.^17,26 A few retrospective (level III/IV) studies support the efficacy and safety of prophylactic IVCF use in elective spine patients.^2,10-13 A survey of spine surgeons suggested they are used on a case-by-case basis.⁹ This study examined all major spine surgery patients at our institution and reported the rate of VTE events and mortality in patients with and without prophylactic IVCFs.

Our overall cohort of patients had VTE, DVT, and PE rates of 9.9%, 5.0%, and 5.8%, respectively (Table 1). Our protocol does not include routine VTE surveillance. Diagnostic testing for VTE was only initiated for symptomatic patients. Differences in patient population and VTE diagnosis cause a wide variation of reported rates (.3-31%).²⁷ A large systematic review found an aggregate DVT and PE rate of 2.1 and .3%, respectively.⁶ A prospective study of adult spinal deformity surgery reported rates of 2.1 and 2.1%, respectively.⁷ Our rates of DVT and PE are higher than these studies but within the literature reported ranges. The all-cause mortality rate in our cohort was 2.3%, comparable to other cohorts of adult major spinal surgery.^7,13,28

A previous study of 318 major spine reconstruction procedures with two-year follow-up reported overall DVT rates of .3%, symptomatic PE rate of 2.2%, and fatal PE rate of .3.³ Their study stratified by approach and an anterior component (anterior only or combined anterior-posterior) identified a PE rate of 6.1% and DVT rate of 2.4% utilizing SCDs and thigh-length compression stockings without chemoprophylaxis.³ The group without IVCF had a similar DVT rate (2.6% in presented study compared to 2.4%) and the overwhelming majority (92.4%) had an anterior component to their approach. The PE rate without IVCF was lower than previously reported (4.6% in presented study compared to 6.1%).³ In other words, the two studies with similar surgical approaches to the thoracolumbar spine determined similar rates of hypercoaguable diagnoses without IVCF. These rates of hypercoagulable diagnoses create interesting avenues for future research to determine why PE and DVT rates of 8.6% were identified in the presented study. The study cohorts risk factors independent of IVCF are an important consideration to standardize in future investigation in order to determine the true effect on overall rates of an IVCF on VTE.

Our institution has no prophylactic IVCF protocol, placed only by surgeon discretion. Overall, 40.7% of the surgical cohort had prophylactic IVCF placement (Table 2). In our study, patients with IVCF were more likely older, obese, undergo anterior spinal fusion, staged surgery, revision surgery, and have preoperative risk factors. Advanced age and obesity have previously been associated with increased VTE risk in orthopaedic surgery.^12,29 Several studies have associated anterior spinal surgery and staged surgery with VTE,^2-5 likely due to manipulation of the retroperitoneal vessels. There was no association between IVCF use and length of the fusion construct, need for osteotomy, duration of surgery, or EBL. Interestingly, despite theoretical concerns of hematoma from chemoprophylaxis, IVCF patients received chemoprophylaxis within 24 hours. There was a significantly higher rate of hematoma/seroma amongst patients with IVCFs, possibly a reflection of the higher use of anterior approach in IVCF patients. IVCF use was positively associated with postoperative DVT, but not associated with PE.

Surgeons fixate particularly on PE due to associated morbidity, but this was rare (15 PE events). Given the scarcity of PE, we chose to study symptomatic VTE (DVT/PE). Based on our univariate analysis, risk factors for postoperative VTE included age, number of levels fused, duration of surgery, blood loss, and known VTE risk factor (Table 3). Prophylactic IVCF use was significantly associated with VTE, particularly with longer indwelling duration (>60 days), reflecting the association between IVCF and risk of DVT development.^30,31 Interestingly, there was a positive association between greater EBL and VTE, but not between TXA use and VTE. This finding suggests TXA may limit blood loss without increasing VTE risk. Interestingly, BMI and anterior spinal fusion were not associated with increased VTE in our study, despite being recognized as potential risk factors in other studies.^2-5,12 Administration of chemoprophylaxis within 24 hours of surgery did not significantly protect against VTE in our univariate analysis.

In the multivariate regression model, significant independent VTE risk factors included age >65 and longer duration of IVCF use (>60 days). Longer duration of IVCF use is confounded by the presence of thrombosis, as a retrievable IVCF would likely remain in place in the setting of DVT. In the multivariate regression model, a prophylactic IVCF was not a significant risk factor for VTE. EBL greater than 1000 mL was approaching significance as an independent risk factor, and along with TXA warrants consideration in future studies.

Complications directly associated with an IVCF are more common than once thought.^14,16,17 One study of 464 patients with IVCF reported a 22% complication rate.¹⁷ Commonly reported complications include inferior vena cava wall penetration, thrombotic occlusion, embolization, IVCF movement or fracture, and access site complications.¹⁶ Previous evidence supported that risk increased with duration of IVCF placement,¹⁵ leading to a shift from permanent to retrievable IVCF designs.¹⁴ Most IVCFs in our study were retrievable designs (94.2%). This decision was decided between the patient and physician placing the device, not the attending spine surgeon. Our IVCF retrieval rate (84.8%) is significantly higher than many rates reported in the literature (12-45%).³² Our IVCF-associated complication rate (10.5%) compares favorably with recent literature.¹⁷ Complications are listed by patient in Table 5.

There are several limitations to our study. The retrospective design causes a lack of randomization (i.e., no prospective control over who received an IVCF) and inherent susceptibility to bias. While IVCF placement varied by surgeon, it was typically due to one or multiple aspects of the patient’s history (Table 2), which creates the possibility of confounding variables contributing to the reported results. Second, retrospective chart reviews are dependent upon manual extraction as well as accurate reporting and coding by the treatment team for complete data collection. Some patients may have received postoperative care outside our hospital system, causing VTE diagnoses to be missed. There is a possibility of bias for patients and physicians to underreport complications, in particular relatively benign or nonspecific complications, from IVCF use such as lymphedema. The outcomes of interest (VTE, DVT, PE, death) were rare in this group of patients, making statistically significant differences challenging to detect. The incidence of VTE was likely underreported, as VTE testing was performed only diagnostically in the setting of symptoms and never for routine asymptomatic surveillance. Finally, we did not address the additional cost of the IVCF implant, its placement, removal, or costs associated with treatment of associated complications.

Conclusion

This study demonstrates that prophylactic IVCF use, specifically duration, and age >65 was associated with an increased risk of symptomatic VTE. Our IVCF related complication rate was relatively low compared to published values, but there were still a clinically significant number of events reported, including one life-threatening event. These conclusions call into question the risk-benefit ratio of routine prophylactic IVCF use and should prompt further study. Given the results of this study performed by an independent observer, the senior author (V.A.) has discontinued prophylactic IVCF use, including in staged anterior and posterior spine surgery.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Ethical Approval

All data was collected retrospectively under IRB approved protocol (above) at a single academic institution that did not require patient informed consent for data collection. IRB Approval – University of Pennsylvania - IRB # 825753; Principal Investigator: Vincent Arlet, MD; Protocol #: 825753; Review Board: IRB #7; University of Pennsylvania, Office of Regulatory Affairs, 3624, Market Street, Suite 301 S, Philadelphia, PA 19104-6006, Ph: 215-573-2540/Fax: 215-573-9438, INSTITUTIONAL REVIEW BOARD, (Federalwide Assurance # 00004028).

ORCID iD

Comron Saifi

References

Geerts

Pineo

Heit

, et al. Prevention of venous thromboembolism: the seventh ACCP conference on antithrombotic and thrombolytic therapy. Chest. 2004;126:338S-400S.

Leon

Rodriguez

Tawk

, et al. The prophylactic use of inferior vena cava filters in patients undergoing high-risk spinal surgery. Ann Vasc Surg. 2005;19:442-447.

Dearborn

Tribus

Bradford

. Thromboembolic complications after major thoracolumbar spine surgery. Spine (Phila Pa 1976). 1999;24:1471-1476.

Ferree

Stern

Jolson

Roberts

Kahn

3rd . Deep venous thrombosis after spinal surgery. Spine (Phila Pa 1976). 1993;18:315-319.

Ferree

Wright

. Deep venous thrombosis following posterior lumbar spinal surgery. Spine (Phila Pa 1976). 1993;18:1079-1082.

Glotzbecker

Bono

Wood

Harris

. Thromboembolic disease in spinal surgery: a systematic review. Spine (Phila Pa 1976). 2009;34:291-303.

Smith

Klineberg

Lafage

, et al. Prospective multicenter assessment of perioperative and minimum 2-year postoperative complication rates associated with adult spinal deformity surgery. J Neurosurg Spine. 2016;25:1-14.

Cain

Jr Major

Lauerman

West

Wood

Fueredi

. The morbidity of heparin therapy after development of pulmonary embolus in patients undergoing thoracolumbar or lumbar spinal fusion. Spine (Phila Pa 1976). 1995;20:1600-1603.

Glotzbecker

Bono

Harris

Brick

Heary

Wood

. Surgeon practices regarding postoperative thromboembolic prophylaxis after high-risk spinal surgery. Spine (Phila Pa 1976). 2008;33:2915-2921.

10.

Rosner

Kuklo

Tawk

Moquin

Ondra

. Prophylactic placement of an inferior vena cava filter in high-risk patients undergoing spinal reconstruction. Neurosurg Focus. 2004;17:E6.

11.

Ozturk

Ganiyusufoglu

Alanay

Aydogan

Onat

Hamzaoglu

. Efficacy of prophylactic placement of inferior vena cava filter in patients undergoing spinal surgery. Spine (Phila Pa 1976). 2010;35:1893-1896.

12.

Dazley

Wain

Vellinga

Cohen

Agulnick

. Prophylactic inferior vena cava filters prevent pulmonary embolisms in high-risk patients undergoing major spinal surgery. J Spinal Disord Tech. 2012;25:190-195.

13.

McClendon

Jr O'shaughnessy

Smith

, et al. Comprehensive assessment of prophylactic preoperative inferior vena cava filters for major spinal reconstruction in adults. Spine (Phila Pa 1976). 2012;37:1122-1129.

14.

Grewal

Chamarthy

Kalva

. Complications of inferior vena cava filters. Cardiovasc Diagn Ther. 2016;6:632-641.

15.

Al-Hakim

Kee

Olinger

Lee

Moriarty

McWilliams

. Inferior vena cava filter retrieval: effectiveness and complications of routine and advanced techniques. J Vasc Intervent Radiol. 2014;25(9):933-940. quiz 940

16.

Duffett

Carrier

. Inferior vena cava filters. J Thromb Haemost. 2017;15:3-12.

17.

Wassef

Lim

. Indications, complications and outcomes of inferior vena cava filters: a retrospective study. Thromb Res. 2017;153:123-128.

18.

Scurr

. How long after surgery does the risk of thromboembolism persist? Acta Chir Scand Suppl. 1990;556:22-24.

19.

Gerlach

Raabe

Beck

Woszczyk

Seifert

. Postoperative nadroparin administration for prophylaxis of thromboembolic events is not associated with an increased risk of hemorrhage after spinal surgery. Eur Spine J. 2004;13:9-13.

20.

Yoon

Kim

Shin

. Postoperative spinal epidural hematoma: risk factor and clinical outcome. Yonsei Med J. 2006;47:326-332.

21.

Kou

Fischgrund

Biddinger

Herkowitz

. Risk factors for spinal epidural hematoma after spinal surgery. Spine (Phila Pa 1976). 2002;27:1670-1673.

22.

Sansone

del Rio

Anderson

. The prevalence of and specific risk factors for venous thromboembolic disease following elective spine surgery. J Bone Joint Surg Am. 2010;92:304-313.

23.

Awad

Kebaish

Donigan

Cohen

Kostuik

. Analysis of the risk factors for the development of post-operative spinal epidural haematoma. J Bone Joint Surg Br. 2005;87:1248-1252.

24.

Uribe

Moza

Jimenez

Green

Levi

. Delayed postoperative spinal epidural hematomas. Spine J. 2003;3:125-129.

25.

Kaufman

Kinney

Streiff

, et al. Guidelines for the use of retrievable and convertible vena cava filters: report from the society of interventional radiology multidisciplinary consensus conference. J Vasc Intervent Radiol. 2006;17:449-459.

26.

Stein

Kayali

Olson

. Twenty-one-year trends in the use of inferior vena cava filters. Arch Intern Med. 2004;164:1541-1545.

27.

Cox

Weaver

Neal

Jacob

Hoh

. Decreased incidence of venous thromboembolism after spine surgery with early multimodal prophylaxis: clinical article. J Neurosurg Spine. 2014;21:677-684.

28.

Kostuik

. Anterior spinal cord decompression for lesions of the thoracic and lumbar spine, techniques, new methods of internal fixation results. Spine (Phila Pa 1976). 1983;8:512-531.

29.

Heck

Brown

Richardson

. Venous thromboembolism in spine surgery. J Am Acad Orthop Surg. 2008;16:656-664.

30.

Decousus

Leizorovicz

Parent

, et al. A clinical trial of vena caval filters in the prevention of pulmonary embolism in patients with proximal deep-vein thrombosis. prévention du risque d’embolie pulmonaire par interruption cave study group. N Engl J Med. 1998;338:409-415.

31.

Greenfield

. The PREPIC study group. Eight-year follow-up of patients with permanent vena cava filters in the prevention of pulmonary embolism: the PREPIC (prevention du risque d’embolie pulmonaire par interruption cave) randomized study. Perspect Vasc Surg Endovasc Ther. 2006;18:187-188.

32.

Angel

Tapson

Galgon

Restrepo

Kaufman

. Systematic review of the use of retrievable inferior vena cava filters. J Vasc Intervent Radiol. 2011;22:1522-1530.