Changes in inferior vena cava filter placement over the past decade at a large community-based academic health center
Article Outline
Objective
A significant increase in the frequency of inferior vena cava (IVC) filter placement at our large community-based academic health center led us to evaluate changes in indications, devices, and providers over the past decade.
Methods
A single-center retrospective review of all filter placements was performed comparing 76 patients in 1995 with 470 patients in 2005. Demographic data, provider data, filter type, and indications for placement were tabulated. Complications, follow-up evaluation, filter removal, and patient outcomes were examined.
Results
There was a greater than sixfold increase in the number of filters placed in 2005 vs 1995. There were no significant differences in patient demographics or the extent of venous thromboembolic (VTE) disease during this period except for an increase in median age. Filter placement by interventional radiologists remained approximately 50% of the total whereas placement by vascular/trauma surgeons increased to 24% and placement by cardiologists decreased to 29% (P < .001). In 2005, a smaller percentage of filters were placed for absolute indications, while filter placements for relative and prophylactic indications increased over the same time period, especially among cardiologists (P = .02). Potentially retrievable filters are increasingly being used for prophylaxis; however, only 2.4% were retrieved. An increasing number of filters were placed in patients with only infrapopliteal deep venous thrombosis (P = .07). A shift was seen to lower profile and removable filter types. Long-term patient follow-up showed little change in disease progression or in morbidity and mortality of filter insertion.
Conclusions
Technological and practice pattern changes have led to an increase in filters inserted by vascular and trauma surgeons in the operating room and intensive care units. Increased diagnosis of VTE disease and newer low profile delivery systems in patients may also have contributed to the significant increase in filter placement. A shift in indications for placement from absolute toward relative indications and prophylaxis is evident over time and across providers, indicating the need for consensus development of appropriate criteria.
The incidence of venous thromboembolism (VTE) has remained relatively constant since about 1980, with an average annual incidence of more than one case per 1000 person years.1 However, the use of inferior vena cava (IVC) filters has increased markedly over the last two decades.2, 3 This may be due to some combination of improved filter technology, liberalization of indications for insertion, an increasing number of providers available for IVC filter placement, and an increased appreciation of the morbid potential of VTE. The changing trends in indications and providers over the last decade caused us to evaluate the experience with IVC filter placement at William Beaumont Hospital, a community-based Academic Health Center located in Royal Oak, Michigan.
Materials and methods
After obtaining approval by the human investigation committee of the hospital, a single-center retrospective study was performed. All IVC filters placed during calendar years 1995 and 2005 at our 1067 bed hospital in Southeastern Michigan were reviewed, comparing a cohort of 76 patients in 1995 with a cohort of 470 patients in 2005. Medical charts and computerized records for each patient were reviewed by physicians. Standard demographic data, provider discipline, filter type, indications for placement, follow-up studies performed, and patient outcomes were collected. The indications for IVC filter placement were grouped as absolute, relative or prophylactic based upon previously published recommendations.4, 5, 6, 7, 8, 9 Patients with documented VTE disease were categorized as having either absolute or relative indications, while patients at high risk but without any documented VTE disease were assigned to the prophylactic category (Table I). Patients having more than one indication were categorized by the most clinically relevant indication. If a filter was placed for a recurrent VTE event despite anticoagulation, the adequacy of anticoagulation was confirmed by reviewing the patient’s prothrombin time, international normalized ratio, and/or activated clotting time values. Anticoagulation was in the majority of cases established with intravenous unfractionated heparin and/or warfarin; a small number of patients in the 2005 group were anticoagulated with subcutaneous low molecular weight heparin.
Table I. Indications for IVC filter placement
| Absolute indications for IVC filter placement in VTE disease: |
| 1. Contraindication to anticoagulation: |
 a. Bleeding complication of anticoagulation or recent history of bleeding that precludes anticoagulation including the following: gastrointestinal or intracranial bleeding, significant hematoma, hemoptysis, hematuria, etc. |
| b. Central nervous system infarct, neoplasm, trauma, or recent/planned CNS surgery |
| c. Significant thrombocytopenia (<50,000/mm3) |
| d. Major extremity/torso trauma precluding anticoagulation, such as due to solid organ injury. |
| 2. Failure of anticoagulation: |
| a. Recurrent pulmonary embolism despite anticoagulation |
| b. Progression of ileo femoral clot despite anticoagulation |
| 3. Heparin associated thrombocytopenia-thrombosis syndrome |
| Relative indications for IVC filter placement in VTE disease: |
| 1. Poor candidate for anticoagulation: |
| a. Old age with risk of falls/ ataxia/ history of seizures |
| b. Poor compliance with anticoagulation medication by patients |
| c. Neoplasm with potential bleeding risk such as GI, GU cancer |
| d. Patient in a periprocedural period having risk of bleeding from anticoagulation |
| 2. Massive PE in which recurrent emboli may prove fatal |
| 3. Ileo femoral deep venous thrombus with a free floating tip |
| 4. During or after surgical embolectomy |
| Prophylactic indications for IVC filter placement with no active VTE disease: |
| 1. Major trauma: |
| a. Multiple long-bone or complex pelvic fractures |
| b. Spinal cord injury with immobilization from para- or quadriplegia |
| 2. Morbidly obese and/or immobile patients |
| 3. High risk patients undergoing: |
| a. Spine surgery |
| b. Bariatric surgery |
Computerized electronic records or paper patient charts were reviewed for recurrent VTE disease as of July 2007. Follow-up studies included lower-extremity venous duplex exams, chest computed tomography (CT) scans performed with a pulmonary embolism (PE) protocol and ventilation-perfusion scans. A follow-up duplex scan was considered positive for worsening deep venous thrombosis (DVT) if it displayed clot progression beyond the baseline study obtained prior to filter placement. To be considered negative for progression of DVT disease, a duplex scan performed at least 30 days after filter placement must show no progression from baseline. Occurrence of a new PE was documented by the presence of new pulmonary artery filling defects on chest CT using high resolution PE protocol. Such studies were obtained based solely on clinical suspicion. Ventilation perfusion scans were used less frequently in 2005; only one patient had a recurrent PE diagnosed with this modality. All reported deaths were reviewed for cause of death and time from placement of filter.
In 1995, IVC filters were placed by cardiologists and radiologists in their angiography suites; at that time, surgeons at our institution did not place filters. In 2005, filters were being placed by cardiologists and radiologists in angiography suites and by vascular and critical care/trauma surgeons, either in the operating room with angiographic capability or in the intensive care with bedside portable fluoroscopy units. In all cases, percutaneous technique was used to gain venous access and venography performed to delineate the venous anatomy and rule out caval thrombosis. Post deployment venography was not always performed; however, in most cases, spot fluoroscopic images were taken to confirm filter alignment and deployment.
Statistical analysis
All categorical variables are shown as count and percent frequencies, and were examined using Pearson χ2 test or Fisher exact test as appropriate. Age is shown as median with 25th to 75th percentile ranges and was examined using a Wilcoxon rank test as age was not normally distributed. In all instances P ≤ .05 was considered significant. SAS version 9.1.3 (Cary, NC) was used for all analyses.
Results
Patient demographics
The absolute number of IVC filters placed increased dramatically from 76 in 1995 to 470 filters in 2005, representing a greater than sixfold increase. During this time, hospital admissions increased 33% (43,770 to 58,106) and orthopedic admissions increased from 8.4% of total admissions to 10.9% (3666 to 6329). Trauma admissions during this time period increased 19% from 1371 to 1624. Median (25th to 75th percentile range) patient age was 68 years (59 to 75 years) in 1995 and 74 years (58 to 83 years) in 2005 (P = .022); in 1995, 47% of patients were male vs 46% in 2005 (P = .860). The median age of patients of all providers increased from 1995 to 2005 (P =.02). Patients with filters placed by critical-care/trauma surgeons in 2005 were significantly younger than patients for other providers. They had a median age of 52 years (38 to 75 years) compared with a median age of 75 years (62 to 83 years) for other providers, reflecting the typically younger trauma patient population (P < .0001).
Associated complications and comorbidities for patient groups were collected and analyzed using ICD-M codes which summarize events recognized during the hospital admission when the IVC filter was placed. Comorbidities are listed in Table II in the online appendix to this publication, which shows respiratory, musculo-skeletal, blood and digestive system comorbidities increased significantly in the 2005 group compared with 1995 patients.
Venous thromboembolic disease
The type of VTE disease in patients undergoing IVC filter placement was comparable in both study periods (P = .23). The majority of filters were placed in patients with documented venous thromboembolic disease; however, increased prophylactic placement of filters in patients at risk for but with no documented VTE disease is seen (Fig 1).
Fig 1.
Venous thromboembolic disease at the time IVC filter placement in 1995 (n = 76) and 2005 (n = 470). No significant difference was seen between the two time periods (P = .23).
Provider distribution
Cardiologists and interventional radiologists placed an equal number of IVC filters in 1995; no filters were placed by surgeons during this period. A decreasing percentage of filters were placed by cardiologists in 2005 vs 1995, while trauma and vascular surgeons accounted for 24% of the total number of filters placed in 2005 (P < .001) (Fig 2).
Fig 2.
Distribution of providers for IVC filter placement in 1995 and 2005.
Indications for placement
A smaller percentage of filters were placed for absolute indications in 2005 vs 1995, while filter placements for relative and prophylactic indications increased over the same time period (P =.12) (Fig 3).
Fig 3.
Indications for IVC filter placement in 1995 and 2005. Differences between the two time periods were not statistically significant (P = .12). N for each group is indicated at the base of each bar.
The majority of the filters placed by cardiologists in 1995 were for absolute indications, while in 2005 the majority of filters placed by cardiologists were for relative and prophylactic indications (P =.020) (Fig 4). The number of filters placed for relative indications by interventional radiologists increased from 1995 to 2005; however, the majority of filters placed by these providers remain for absolute indications (P =.350) (Fig 4). A trend change cannot be analyzed for surgeons since they did not place filters in 1995. When all providers are considered, cardiologists place the highest percentage of filters for relative indications and the lowest percentage for absolute indications. The largest percentage of filters placed for prophylactic indications are placed by trauma surgeons (Fig 5).
Fig 4.
Indications for IVC filter placement by cardiologists and radiologists in 1995 and 2005. Statistically significant differences were seen between the two time periods for cardiologists (P = .02) but not for radiologists (P = .35). N for each group is indicated at the base of each bar.
Fig 5.
Distribution of indications for IVC filter placement among all providers during 2005.
The percentage of filters placed for absolute indications decreased from 67% in 1995 to 57% in 2005 but this decrease was not significant (P = .09). There was a significant increase (3.9% in 1995 to 13.5% in 2005, P = .05) in the number of filters inserted to protect the neurosurgical patient population in whom anticoagulation was judged unacceptable. There was a significant decrease in the number of filters placed for recurrent PE despite anticoagulation (31.4% in 1995 to 15% in 2005 P =.0049). The frequency of other absolute indications for filter placement as listed in Table I remain unchanged and are available in detail in Table III in the online appendix for this publication.
The percentage of filters inserted for relative indications was similar in 1995 and in 2005 (30% vs 35%, P = .39). There was a decrease in the percentage of filters used solely for peri-operative protection (47.8% in 1995 to 17.5% in 2005, P =.0021), with concomitant adaptation of anticoagulation protocols. However, there was a significant increase (17.4% in 1995 to 54.8% in 2005, P = .0008) in the number of filters placed in patients considered to be poor candidates for anticoagulation based on old age and/ or risk of falls. This trend is present across all providers, although the increase reaches statistical significance only for filter placement by radiologists (0% in 1995 to 59% in 2005, P =.0002). The frequency of other relative indications for filter placement as listed in Table I remain unchanged and are available in detail in Table IV in the online appendix for this publication.
The number of filters inserted for prophylactic indications rose from 2.6% in 1995 to 7.9% in 2005 (P =.10). Fully 58.6% of prophylactic filters in 2005 were inserted in trauma patients with multiple long bone/pelvic fractures or spinal cord injury and 32.4% of prophylactic filters were in bariatric surgery patients. One patient had an IVC filter inserted with no apparent indication for anticoagulation in the medical records.
Patients with infrapopliteal lower extremity DVT only
For patients with lower extremity DVT confined to infrapopliteal veins only, the placement of IVC filters increased from 2.6% (2/76) in 1995 to 8.5% (40/470) in 2005 (P =.07). The two filters placed in 1995 for infrapopliteal disease were for absolute indications, but in 2005 45% (18/40) of filters placed in patients with infrapopliteal DVT were placed for the relative indication of the patient as a poor candidate for anticoagulation due to old age/risk of falls (13/40, 33%), neoplasm with potential for bleeding (2/40, 5%) or proximity of other procedures (3/40, 8%). Table V in the online appendix to this publication shows in detail the number of filters placed in such patients for absolute and relative indications by provider type. Follow-up imaging and outcomes in this group of patients is reported below.
Filter type and the retrieval of filters
The types of IVC filters placed in 1995 and 2005 are shown in Table VI. The majority of filters placed in 1995 were either the titanium or stainless steel percutaneous Greenfield type. In 2005, a greater variety of filters were placed, with a shift to the more recently available low profile and potentially retrievable filters (<7F). The use of the 12F Greenfield filter (Boston Scientific, Natick, Mass) decreased from 83% of all filters placed in 1995 to 8% in 2005. Retrievable filters were increasingly used (n = 167) but only four (2.4%) were retrieved. The most common indication at the time of placement for retrievable filters was an absolute contraindication to anticoagulation (34.1%); this is followed by patients who are poor candidates for anticoagulation due to old age and/or risk of falls (12%) (Table VII). The providers for placement and retrieval of these filters are shown in Table VIII. One additional patient underwent an unsuccessful attempt at retrieval 4 months after placement.
Table VI. Type of IVC filters placed in 1995 and 2005, including number of retrievable filters retrieved
| Filter type | 1995 N = 76 | 2005 N = 470 | Number of retrievable filters retrieved |
|---|---|---|---|
| Greenfield (titanium/stainless steel) (12F) | 63(83%) | 38(8%) | |
| Vena Tech (LGM/LP) (12F/7F) | 10(13%) | 116(25%) | |
| Gunther Tulip (9F – retrievable) | 0 | 107(23%) | 2 |
| Trap-Ease (6F) | 0 | 141(30%) | |
| Opt-Ease (6F – retrievable) | 0 | 60(13%) | 2 |
| Simon-Nitinol (9F) | 2(3%) | 2(0.4%) | |
| Bird’s Nest (14F) | 1(1%) | 0 |
Table VII. Subindications for filter placement at the time of retrievable filter insertion
| Filter indications at the time of retrievable filter placement | Number of retrievable filters placed |
|---|---|
| Absolute - bleeding complication | 57(34.1%) |
| Absolute – CNS infarct/tumor/trauma surgery | 16(9.6%) |
| Absolute – thrombocytopenia | 3(1.8%) |
| Absolute – trauma – solid organ injury | 5(3.0%) |
| Absolute – recurrent PE with anticoagulation | 12(7.2%) |
| Absolute – progression of DVT with anticoagulation | 3(1.8%) |
| Absolute – heparin associated thrombocytopenia | 0(0%) |
| Relative – poor candidate – old age/falls | 20(12.0%) |
| Relative – poor compliance | 4(2.4%) |
| Relative – neoplasm with potential to bleed | 2(1.2%) |
| Relative – periprocedural period | 13(7.8%) |
| Relative – massive PE | 5(3.0%) |
| Relative – free-floating ileo femoral DVT tip | 3(1.8%) |
| Relative – during/after surgical embolectomy | 0(0%) |
| Prophylactic – long bone/pelvic fractures | 9(5.4%) |
| Prophylactic – spinal cord injury | 8(4.8%) |
| Prophylactic – morbidly obese/ immobile patient | 4(2.4%) |
| Prophylactic – prior to spine surgery | 1(0.6%) |
| Prophylactic – prior to bariatric surgery | 2(1.2%) |
Table VIII. Providers for retrievable IVC filters
| Provider | Number of retrievable filters inserted | Number of retrievable filters retrieved | Time interval between placement and retrieval |
|---|---|---|---|
| Cardiologist | 2(1.3%) | 1 | 1mo |
| Radiologist | 98(59.0%) | 1 | 1mo |
| Vascular surgeon | 14(8.4%) | 1 | 3wk |
| Trauma surgeon | 52(31.3%) | 1 | 1mo |
Follow-up imaging and patient outcomes
Medical records were reviewed to identify patient outcomes following filter placement. Table IX shows the follow-up imaging and outcomes of the 2005 year patients; it also highlights the percentage of these patients who received post-filter anticoagulation. The majority of these patients were anticoagulated with intravenous unfractionated heparin and oral warfarin. Low molecular weight heparin (Lovenox) using DVT treatment dosage was prescribed in 1.7% (8/470) of these patients. Interestingly, 14 of 91 (15%) patients considered poor candidates for anticoagulation at the time of filter placement due to age/risk of falls continued to receive anticoagulation post-filter placement. A detailed breakdown of follow-up data by subindication for filter placement as listed in Table I is available in Table IX, a in the online appendix for this publication. Follow-up duplex imaging was performed in patients with clinical suspicion of PE or lower extremity clot progression. Lower extremity ultrasound was performed on 28% of these patients (131/470); 49% (64/131) showed clot progression. Similarly, follow-up chest CT scan with PE protocol was performed in 12% (55/470) of patients; 12% (7/55) of these showed new pulmonary embolism. Follow-up duplex imaging in the lower extremity infrapopliteal DVT-only group was performed on 25% (10/40) of these patients; 60% (6/10) revealed DVT clot progression.
Table IX. Post-filter anticoagulation and follow-up imaging for VTE disease in 2005 patient group
| Indication for filter placement | N | Post-filter anticoagulation | Follow-up duplex ultrasound | Follow-up chest CT | ||||
|---|---|---|---|---|---|---|---|---|
| No | Yes – no progression | Yes – clot progression | No | Yes – no new PE | Yes – new PE | |||
| Absolute | 267 | 49(18%) | 187(70%) | 35(13%) | 45(17%) | 231(87%) | 31(12%) | 5(1.9%) |
| Relative | 166 | 58(35%) | 121(73%) | 28(17%) | 17(10%) | 149(90%) | 15(9%) | 2(1.2%) |
| Prophylactic | 37 | 2(5.4%) | 31(84%) | 4(11%) | 2(5.4%) | 35(95%) | 2(5.4%) | 0 |
| Total | 470 | 109(23%) | 339(72%) | 67(14%) | 64(14%) | 415(88%) | 48(10%) | 7(2%) |
| Lower extremity infrapopliteal DVT only | 40 | 5(12.5%) | 30(75%) | 4(10%) | 6(15%) | 35(88%) | 5(12.5%) | 0 |
Complications
The complications observed in 1995 patient group included a neck hematoma (1/76, 1.3%), respiratory insufficiency within 24 hours of filter placement (2/76, 2.6%), clot progression on follow-up duplex scan (2/76, 2.6%), death from cardio-pulmonary failure, and/or sepsis within a week of filter placement (3/76, 3.9%). The complications observed in the 2005 patient group included groin hematoma (2/470, 0.4%), phlegmasia cerulea dolens at 10 days (1/470, 0.2%) and 3 weeks (1/470, 0.2%) following filer placement, and IVC thrombosis (5/470, 1.1%; two requiring thrombectomy). It is not possible to determine if these complications result from filter insertion or represent progression of the original disease state. The features of the hospital course for the 2005 year patients who developed caval thrombosis or phlegmasia cerulea dolens are shown in Table X in the online appendix to this publication.
Table X. Features of patients complicated by caval thrombosis or phlegmasia cerulea dolens after IVC filter placement in 2005
| Indication for filter | Provider | Time from filter insertion | Post-filter anti-coagulation | Patient outcome | Management of complication | |
|---|---|---|---|---|---|---|
| Pt 1 | Relative - poor compliance with anticoagulation | Radiologist | 2 wk | No | Caval thrombosis and PE | Placement of second filter above the first by radiologist |
| Pt 2 | Relative - colonic neoplasm with potential to bleed on anticoagulation | Cardiologist | 2 wk | No | Caval thrombosis | IVC thrombectomy by radiologist |
| Pt 3 | Absolute - contraindication to anticoagulation due to hemorrhagic pericardial effusion | Cardiologist | 3 wk | No | Caval thrombosis | Placement of suprarenal filter above the first by vascular surgeon |
| Pt 4 | Absolute - recurrent PE despite anticoagulation | Radiologist | 1 y | No | Caval Thrombosis and PE | IVC/iliac vein thrombectomy then stent placement by cardiologist |
| Pt 5 | Absolute – liver laceration as contraindication to anticoagulation for PE+DVT | Trauma surgeon | 2 wk | No | Caval thrombosis | IVC Thrombectomy by radiologist |
| Pt 6 | Absolute - recurrent PE despite anticoagulation | Cardiologist | 3 wk | Yes | Phlegmasia cerulea dolens and venous gangrene of lower extremity | Patient made hospice and died of sepsis/metastatic cancer |
| Pt 7 | Absolute - recurrent PE despite anticoagulation | Radiologist | 10 d | Yes | Phlegmasia cerulea dolens and venous gangrene of lower extremity | Patient died of sepsis/stroke. |
Of the 2005 year patients, 81 of 470 (17.2%) died in the 2-year follow-up period (Table XI), and 89% (72/81) of these patients died within 6 months of filter placement. Six deaths occurred within 24 hours of filter placement in 2005. None of these patient deaths were a direct result of technical complications of filter placement. However, one patient died from aspiration pneumonia (1/470, 0.2%) and one from PE (1/470, 0.2%) within 24 hours of filter placement, which may have contributed to their deaths.
Table XI. Time line and cause of death over 2-year follow-up in the 2005 patients
| PE | Aspiration pneumonia | Cardio- respiratory failure | MOSF/ sepsis | Metastatic cancer | CVA/ ICH | Total | |
|---|---|---|---|---|---|---|---|
| < 24 h | 1 | 1 | 0 | 2 | 2 | 0 | 6 |
| < 1 wk | 1 | 1 | 2 | 6 | 5 | 2 | 17 |
| < 1 mo | 1 | 1 | 5 | 12 | 6 | 3 | 28 |
| < 6 mo | 0 | 1 | 3 | 7 | 9 | 1 | 21 |
| < 1 y | 0 | 1 | 1 | 3 | 0 | 1 | 6 |
| < 2 y | 0 | 0 | 0 | 2 | 1 | 0 | 3 |
| Total | 3 | 5 | 11 | 32 | 23 | 7 | 81 |
Discussion
Anticoagulation remains the standard approach for patients with VTE disease. The American College of Chest Physicians (ACCP) consensus conference recommends the utilization of IVC filters in patients with a contraindication to (or a complication of) anticoagulation, heparin-induced thrombocytopenia, and in patients with recurrent VTE disease despite anticoagulation.5 Previous randomized controlled trials have demonstrated the efficacy of prophylactic placement of IVC filters to reduce the short-term risk of pulmonary embolism in patients with documented VTE disease treated with anticoagulation albeit with an increased long-term incidence of recurrent DVT and without any long-term reduction in mortality.10
The present study demonstrates significant changes in the placement of IVC filters at a high-volume center, including a reduction in the proportion of filters being placed for absolute indications from 67.1% in 1995 to 56.8% in 2005, suggesting that a greater percentage of IVC filters are being placed for relative and prophylactic indications.
We observed an increase in the placement of filters for progression of DVT in spite of anticoagulation, which did not reach statistical significance; this may relate to improved technology and ease of venous duplex scanning now available compared with 1995. Our study also demonstrated a significant increase in placement of filters in patients with VTE disease and central nervous system (CNS) infarct, tumor, or planned CNS surgery. There is little long-term data on the outcome of filters in this particular patient group; however, in the present study 13.5% of filters in 2005 were in patients with VTE disease associated with either CNS infarct, tumors, or planned CNS surgery. When clinically indicated, lower extremity duplex ultrasound follow-up in 28% (10/36) of this cohort showed 9/10 (90%) with DVT clot progression. Similarly, follow-up chest CT scan in 8% (3/36) of these patients showed recurrent PE in the absence of anticoagulation in 1/3 (33%). Filter placement for the indication of recurrent PE despite anticoagulation showed a statistically significant decrease. This decrease may have been in part due to the fact that in 1995 the diagnosis of PE was based largely on clinical suspicion with a “suggestive” ventilation-perfusion scan, whereas the diagnosis in 2005 was more objectively based on the presence of filling defects seen on high resolution helical CT scans of the chest. Further study is warranted into the safety, efficacy, and cost-effectiveness of IVC filter placement in this cohort.
Another group for which an increased incidence of IVC filter placement was noted was in elderly patients considered “poor candidates” for anticoagulation owing to comorbidities or perceived risk of falls. Filter placement for this indication has increased from 4/76 (5.3%) filters in 1995 to 91/470 (19.4%) filters in 2005. Elderly patients are known to be at increased for VTE disease with a higher incidence of recurrence.11, 12 Moreover, the risks of anticoagulation, particularly intracranial hemorrhage with relatively minor head injury is increased dramatically in the elderly patient.13, 14, 15 As the population ages, the incidence of VTE disease and the incidence of IVC filter placement are likely to increase in parallel suggesting a clear need for efficacy and outcome studies in this high risk population.
Malignancy is an independent risk factor for VTE.1, 16 IVC filter placement appears to be effective in preventing PE, but there is limited survival benefit in this particular patient population.17 Analysis in this study on cause of death within 2 years of filter placement showed 28% (23/81) of patients who died, died from metastatic cancer, all but one within 6 months of filter placement. Ambulatory status at time of placement is a significant predictor of longer survival,18 possibly allowing identification of a subset of patients with malignancy who will truly benefit from IVC filter placement.
Rutherford argues that there has been an increased use of potentially retrievable IVC filters for prophylactic indications.19 Retrievable IVC filters were not designed for long-term placement and studies are not yet available on their long-term safety. Even low complication rates when multiplied over the lifetime of a young patient could become important and must enter into the decision to place a prophylactic retrievable IVC filter.6 Giannoudis et al recently reviewed all available studies of IVC filter efficacy in trauma patients and concluded that despite a low incidence of PE in trauma patients, this complication has remained a significant cause of death, and that filters appear to be both safe and effective in reducing this complication.20 A survey in 1997 of trauma surgeons found that the potential removability of filters would significantly increase prophylactic filter placement in high-risk patients from 29% to 53%.21 Reported retrieval rates vary from 34% to 84%.22, 23, 24, 25, 26 In the present study, trauma surgeons did not place any IVC filters in 1995, but were responsible for 11.3% of filters in trauma and surgical ICU patients (all of which were retrievable types) by 2005. Interestingly, only 2.4% of potentially retrievable filters were removed in this study. The majority of potentially retrievable filters that were not recovered were placed by radiologists at the request of medical providers and the poor rate of retrieval most likely represents inadequate follow-up by the original provider. The second largest group of potentially retrievable filters was placed in trauma patients, where the potential retrieval of the filter is desirable but may not be possible during the recommended period for retrieval due to the severity of the original injuries. In such trauma cases, where there is no reasonable expectation of filter retrieval, prudence would suggest the use of a filter type for which long-term (permanent) placement has been studied and is accepted. Similar considerations influence the placement of filters in other young age populations such as bariatric surgery patients receiving preoperative prophylactic filters, or pregnant women with DVT. The indications for the placement of prophylactic filters need to be critically evaluated and protocols should be developed to ensure proper follow-up and timely removal of filters.
Despite relative constancy in the incidence of VTE, the absolute number of IVC filter placements increased sixfold over 10 years at our large (1067 bed) community-based academic health center. This occurred despite the fact that total hospital admissions increased by only 33% during this period. A statistically significant increase in filter placement was observed from 76 of 43,770 total hospital admissions in 1995 (0.17%) to 470 of 58,106 admissions in 2005 (0.81%) (P < .0001). Demographic factors that may have contributed to increased placement include designation as a level I trauma center in 1998 and the expansion of orthopedic surgical services with an overall doubling of total joint replacement and spine procedures. Major trauma, total joint replacement, and spine surgery procedures are well documented risk factors for postoperative VTE.1 Orthopedic admissions increased from 3666 of 43,770 in 1995 (8.4%) to 6329 of 58,106 in 2005 (10.9%). This 2% increase in orthopedic admissions does not account for the greater than fourfold increase in filter placements as a percentage of total admissions (P < .0001). Similarly, while overall trauma admissions increased from 1371 of 43,770 in 1995 to 1624 of 58,106 in 2005, trauma admissions decreased as a percentage of the total from 3.1% to 2.8% over this past decade. Therefore, trauma admissions do not fully explain the increased prevalence of filter placement. In addition to demographic and procedural trends, an increased recognition of the prevalence, natural history, and clinical consequences of VTE disease over the past decade may also have favored more frequent filter placement. Improved imaging capabilities likely increased detection of VTE disease and improvements in filter design including lower profile delivery systems have facilitated placement by decreasing the threshold for placement and increasing the therapeutic index.
Proprietary filter design also varied considerably over this study, reflecting new lower profile devices and the desire of clinicians to gain experience with devices. In latter years, the use of potentially retrievable devices has clearly increased. This study did not attempt to address differences in safety and efficacy among IVC filter devices.
Nevertheless, the aforementioned factors do not account for the provider shifts observed during the 10-year period of this study. These appear to be related to both technologic advances and practice pattern changes. Prior to 1995, IVC filters were placed almost exclusively by surgeons because they required a surgical exposure of the internal jugular or common femoral vein. Design modifications of the prototypical stainless steel Greenfield filter facilitating percutaneous placement led to a practice change whereby surgeons initially delegated this procedure to interventional radiologists and subsequently to cardiologists. The evolution of vascular surgical practice, in particular the rapid assimilation of endovascular skills, resulted in a shift back to vascular surgeons and trauma surgeons for filter placement first in the operating room and subsequently in the intensive care unit. Surgeons now place 24% of all IVC filters at our institution.
A key observation of the present study is the apparent loosening of the indications for filter placement. Prophylactic IVC filter placement increased from 2.6% to 7.9% of total filter placements over the past decade indicating perhaps an increased recognition of the potential morbidity of VTE and a reduced procedural morbidity. Increased utilization by trauma surgeons in patients known to be at high risk represents the greatest number and percent increase of prophylactic filters. Combined absolute and relative indications were overwhelmingly (> 90%) the indication for filter placement in both time periods and among all provider groups. However, placement for absolute indications decreased from 67% to 57% while placement for relative indications increased from 25% to 35% (P = .061), perhaps also reflecting the perception of decreased morbidity and ease of placement with the newer low-profile filter designs.
Another key observation in the present study is the marked increase in placement for calf vein (infrapopliteal) DVT from 2.6% in 1995 to 8.5% in 2005. While not statistically significant, this represents a clear and disturbing trend toward loosening in the indications for IVC filter placement. Follow-up duplex imaging in patients in the infrapopliteal DVT group with worsening clinical symptoms following filter insertion (28%, 10/36) documented progression in 90% of these patients (9/10); raising a significant concern for inappropriate and excessive use of filters in this subgroup. Follow-up duplex imaging in patients in the infrapopliteal DVT group with worsening clinical symptoms following filter insertion (28%, 10/36) documented progression in 60% (6/10) of these patients, raising a significant concern for inappropriate and excessive use of filters in this subgroup. Of those patients receiving an IVC filter for infrapopliteal DVT (n = 40), the vast majority were not being treated with any anticoagulation (88%). None of the IVC filters placed for infrapopliteal DVT in patients that were not being treated with anticoagulation were placed by vascular surgeons. Moreover, none of the patients receiving IVC filters for infrapopliteal DVT who had documented progression (n = 6) were being treated with anticoagulants. This suggests a possible misunderstanding on the part of the treating physician with respect to both the indications for IVC filter placement and the natural history of venous thromboembolic disease. Thus, the results of the present study suggest that increased facility of percutaneous placement has increasingly removed the vascular surgeon from the decision-making process for filter placement as demonstrated both by changes in providers and increased placement for relative indications and prophylaxis. Further studies to document the impact of this trend in safety and cost-effectiveness appear to be clearly indicated.
In summary, a number of factors appear to be driving increased placement of IVC filters at our large tertiary care institution: increased recognition and awareness of the consequences of venous thromboembolic disease, increases in the number of trauma patients and patients having high risk orthopedic procedures, increased availability and reliability of noninvasive venous duplex scans, changes in the numbers of providers qualified to place IVC filters, changes in filter design and technology allowing percutaneous placement, and finally, a shift from absolute to relative and prophylactic indications for IVC filter placement. The latter three factors deserve further study to assess in an objective, quantifiable fashion the long-term morbidity related to filter placement and further longitudinal follow-up is needed to assess outcomes in patients receiving filters for relative indications.
Author contributions
Appendix
Additional material for this article may be found online at www.jvascsurg.org
Table II. online only. Patient demographics
| Underlying disease⁎ | No of events (%) 1995 N = 76 | No of events (%) 2005 N = 470 | P value |
|---|---|---|---|
| Diseases of respiratory system | 28(37%) | 238(51%) | .026 |
| Diseases of cardiovascular system | 55(72%) | 359(76%) | .45 |
| Diseases of central nervous system | 19(25%) | 174(37%) | .042 |
| Diseases of musculo-skeletal system | 19(25%) | 190(40%) | .010 |
| Diseases of skin | 13(17%) | 94(20%) | .56 |
| Diseases of endocrine system | 21(28%) | 166(35%) | .19 |
| Diseases of blood | 20(26%) | 221(47%) | .0007 |
| Diseases of digestive system | 27(36%) | 236(50%) | .017 |
| Neoplastic disorder | 23(30%) | 150(32%) | .77 |
| Diseases of genito-urinary system | 45(59%) | 253(54%) | .38 |
| Infectious disorder | 44(58%) | 232(49%) | .17 |
| Injury | 9(12%) | 61(13%) | .78 |
| Complication of pregnancy | 1(1.3%) | 0(0%) | .14 |
⁎Underlying disease at the time of Inferior vena cava (IVC) filter placement based on ICD.9.CM codes for events occurring during the admission for IVC filter placement. |
Table III. online only. Number of IVC filters placed for each indication
| Filters placed for absolute indications | 1995 N = 51/76 (67%) | 2005 N = 267/470 (57%) | P value .09 |
|---|---|---|---|
| Bleeding complication | 30(58.8%) | 170(63.7%) | .51 |
| CNS infarct/tumor/trauma surgery | 2(3.9%) | 36(13.5%) | .05 |
| Thrombocytopenia | 2(3.9%) | 8(3.0%) | .67 |
| Trauma – solid organ injury | 0(0%) | 5(1.9%) | 1.00 |
| Recurrent PE with anticoagulation | 16(31.4%) | 40(15.0%) | .0049 |
| Progression of DVT with anticoagulation | 0(0%) | 5(1.9%) | 1.00 |
| Heparin associated thrombocytopenia | 1(2.0%) | 3(1.1%) | .50 |
| Filters placed for relative indications | 1995 N = 23/76 (30%) | 2005 N = 166/470 (35%) | P value .39 |
|---|---|---|---|
| Poor candidate – old age/ falls | 4(17.4%) | 91(54.8%) | .0008 |
| Poor compliance | 1(4.4%) | 13(7.8%) | 1.00 |
| Neoplasm with potential to bleed | 3(13.0%) | 12(7.2%) | .40 |
| Periprocedural period | 11(47.8%) | 29(17.5%) | .0021 |
| Massive PE | 3(13.0%) | 14(8.4%) | .44 |
| Free-floating ileo femoral DVT tip | 1(4.4%) | 6(3.6%) | 1.00 |
| During/after surgical embolectomy | 0(0%) | 1(0.6%) | 1.00 |
| Filters placed for prophylactic indications | 1995 N = 2/76 (2.6%) | 2005 N = 37/470 (7.9%) | P value .10 |
|---|---|---|---|
| Long bone/pelvic fractures | 2(100%) | 9(24.3%) | .07 |
| Spinal cord injury | 0(0%) | 9(24.3%) | 1.00 |
| Morbidly obese/immobile patient | 0(0%) | 6(16.2%) | 1.00 |
| Prior to spine surgery | 0(0%) | 1(2.7%) | 1.00 |
| Prior to bariatric surgery | 0(0%) | 12(32.4%) | 1.00 |
Table IV. online only. Providers for IVC filters placed with relative indications
| Filters placed by cardiologists for relative indications | 1995 N = 12/38 (32%) | 2005 N = 63/135 (47%) | P value .10 |
|---|---|---|---|
| Poor candidate – old age/falls | 4(33%) | 35(56%) | .16 |
| Poor compliance | 1(8.3%) | 5(7.9%) | 1.00 |
| Neoplasm with potential to bleed | 1(8.3%) | 7(11.1%) | 1.00 |
| Periprocedural period | 4(33%) | 11(17.5%) | .24 |
| Massive PE | 1(8.3%) | 4(6.4%) | 1.00 |
| Free-floating ileo femoral DVT tip | 1(8.3%) | 1(1.6%) | .30 |
| During/after surgical embolectomy | 0(0%) | 0(0%) | *** |
| Filters placed by radiologists for relative indications | 1995 N = 11/38 (29%) | 2005 N = 78/223 (35%) | P value .47 |
|---|---|---|---|
| Poor candidate – old age/falls | 0(0%) | 46(59%) | .0002 |
| Poor compliance | 0(0%) | 8(10%) | .59 |
| Neoplasm with potential to bleed | 2(18%) | 5(6.4%) | .21 |
| Periprocedural period | 7(64%) | 8(10%) | .0002 |
| Massive PE | 2(18%) | 7(9%) | .31 |
| Free-floating ileo femoral DVT tip | 0(0%) | 3(3.9%) | 1.00 |
| During/after surgical embolectomy | 0(0%) | 1(1.3%) | 1.00 |
| Filters placed by surgeons in 2005 for relative indications | Trauma surgeons N = 9 | Vascular surgeons N =16 |
|---|---|---|
| Poor candidate – old age/falls | 1(11%) | 9(56%) |
| Poor compliance | 0(0%) | 0(0%) |
| Neoplasm with potential to bleed | 0(0%) | 0(0%) |
| Periprocedural period | 4(44%) | 6(37.5%) |
| Massive PE | 2(22%) | 1(6.3%) |
| Free-floating ileo femoral DVT tip | 2(22%) | 0(0%) |
| During/after surgical embolectomy | 0(0%) | 0(0%) |
Table V. online only. Provider and indications for IVC filter placement in patients with infrapopliteal DVT only
| Indication for placement | Cardiologist | Radiologist | Vascular surgeon | Trauma surgeon |
|---|---|---|---|---|
| Absolute – bleeding complication | 7(38.9%) | 8(38.1%) | 0 | 0 |
| Absolute – CNS infarct/tumor/trauma surgery | 1(5.6%) | 3(14.3%) | 0 | 0 |
| Absolute – recurrent PE with anti-coagulation | 0 | 1(4.8%) | 0 | 0 |
| Absolute – heparin associated thrombocytopenia | 1(5.6%) | 1(4.8%) | 0 | 0 |
| Relative – poor candidate old age/falls | 6(33.3%) | 6(28.6%) | 1 | 0 |
| Relative – neoplasm with potential to bleed | 2(11.1%) | 0 | 0 | 0 |
| Relative – periprocedural period | 1(5.6%) | 2(9.5%) | 0 | 0 |
| Total = 40 in 2005 | 18 | 21 | 1 | 0 |
Table IX, a. online only. Post-filter anticoagulation and follow-up imaging for VTE disease in 2005 patient group with breakdown by indication subtype
| Filters placed for absolute Indications | N | Post-filter anticoagulation | Follow-up duplex ultrasound | Follow-up chest CT | ||||
|---|---|---|---|---|---|---|---|---|
| No | Yes – no progression | Yes – clot progression | No | Yes – no new PE | Yes – new PE | |||
| Bleeding complication | 170 | 13(7.7%) | 122(72%) | 20(12%) | 28(16%) | 148(87%) | 19(11%) | 3(1.7%) |
| CNS infarct/tumor/trauma surgery | 36 | 5(13.9%) | 26(72%) | 1(2.8%) | 9(25%) | 33(92%) | 2(5.6%) | 1(2.8%) |
| Thrombocytopenia | 8 | 3(37.5%) | 5(63%) | 2(25%) | 1(13%) | 8(100%) | 0 | 0 |
| Trauma – solid organ injury | 5 | 2(40%) | 3(60%) | 1(20%) | 1(20%) | 6(100%) | 0 | 0 |
| Recurrent PE with anticoagulation | 40 | 23(57.5%) | 24(60%) | 11(28%) | 59(12%) | 29(73%) | 10(25%) | 1(2.5%) |
| Progression of DVT with anticoagulation | 5 | 2(40%) | 4(80%) | 0 | 1(20%) | 5(100%) | 0 | 0 |
| Heparin associated thrombocytopenia | 3 | 1(33%) | 3(100%) | 0 | 0 | 3(100%) | 0 | 0 |
| Totals | 267 | 49(18%) | 187(70%) | 35(13%) | 45(17%) | 231(87%) | 31(12%) | 5(1.9%) |
| Filters placed for relative indications | N | Post-filter anticoagulation | Follow-up duplex ultrasound | Follow-up chest CT | ||||
|---|---|---|---|---|---|---|---|---|
| No | Yes – no progression | Yes – clot progression | No | Yes – no new PE | Yes – new PE | |||
| Poor candidate – old age/ falls | 91 | 14(15%) | 75(82%) | 11(12%) | 5(5.5%) | 87(96%) | 4(4%) | 0 |
| Poor compliance | 13 | 5(38%) | 8(62%) | 2(15%) | 3(23%) | 8(62%) | 3(23%) | 2(15%) |
| Neoplasm with potential to bleed | 12 | 5(42%) | 7(58%) | 2(17%) | 3(25%) | 11(92%) | 1(8.3%) | 0 |
| Periprocedural period | 29 | 17(59%) | 18(62%) | 7(24%) | 4(14%) | 27(93%) | 2(6.8%) | 0 |
| Massive PE | 14 | 12(86%) | 8(57%) | 4(29%) | 2(14%) | 10(71%) | 4(29%) | 0 |
| Free-floating ileo femoral DVT tip | 6 | 4(67%) | 4(67%) | 2(33%) | 0 | 5(83%) | 1(17%) | 0 |
| During/ after embolectomy | 1 | 1(100%) | 1(100%) | 0 | 0 | 1(100%) | 0 | 0 |
| Totals | 166 | 58(35%) | 121(73%) | 28(17%) | 17(10%) | 149(90%) | 15(9%) | 2(1.2%) |
| Filters placed for prophylactic indications | N | Post-filter anticoagulation | Follow-up duplex ultrasound | Follow-up chest CT | ||||
|---|---|---|---|---|---|---|---|---|
| No | Yes – no progression | Yes– clot progression | No | Yes – no new PE | Yes – new PE | |||
| Long bone/ pelvic fractures | 9 | 0(0%) | 7(78%) | 2(22%) | 0 | 9(100%) | 0 | 0 |
| Spinal cord injury | 9 | 1(11%) | 7(78%) | 1(11%) | 1(11%) | 7(78%) | 2(22%) | 0 |
| Morbidly obese/ immobile patient | 6 | 0(0%) | 5(83%) | 1(17%) | 0 | 6(100%) | 0 | 0 |
| Prior to spine surgery | 1 | 1(100%) | 1(100%) | 0 | 0 | 1(100%) | 0 | 0 |
| Prior to bariatric surgery | 12 | 0(0%) | 11(92%) | 0 | 1(8%) | 12(100%) | 0 | 0 |
| Totals | 37 | 2(5.4%) | 31(84%) | 4(11%) | 2(5.4%) | 35(95%) | 2(5.4%) | 0 |
| Patients with lower extremity infrapopliteal DVT only | 40 | 5(12.5%) | 30(75%) | 4(10%) | 6(15%) | 35(88%) | 5(12.5%) | 0 |
References
- . The epidemiology of venous thromboembolism in the community: implications for prevention and management. J Thromb Thrombolysis. 2006;21:23–29
- . Twenty-one-year trends in the use of inferior vena cava filters. Arch Intern Med. 2004;164:1541–1545
- . Inferior vena caval filters: review of a 26-year single-center clinical experience. Radiology. 2000;216:54–66
- Quality improvement guidelines for percutaneous permanent inferior vena cava filter placement for the prevention of pulmonary embolism. J Vasc Interv Radiol. 2003;14:S271–S275
- . Antithrombotic therapy for venous thromboembolic disease: the Seventh ACCP Conference on Antithrombotic and Thrombolytic Therapy. Chest. 2004;126(3 Suppl):401S–428S
- Practice Parameter Workgroup for DVT Prophylaxis. Practice management guidelines for the management of venous thromboembolism in trauma patients [Manual on the Internet]. Eastern Association for the Surgery of Trauma (EAST), East Northport, NY [cited 2006 Sept 5]. Available from: http://www.east.org/tpg/dvt.pdf.
- . The role of vena caval filters in the management of venous thromboembolism. Blood Rev. 2005;19:179–202
- . Update on inferior vena cava filters. J Vasc Interv Radiol. 2003;14:425–440
- . Inferior vena caval filters: key considerations. Am J Med Sci. 2005;330:82–87
- A clinical trial of vena caval filters in the prevention of pulmonary embolism in patients with proximal deep-vein thrombosis (Prevention du Risque d’Embolie Pulmonaire par Interruption Cave Study Group). N Engl J Med. 1998;338:409–415
- . Management of venous thromboembolism in the elderly. Drugs Aging. 2006;23:651–671
- . Frailty and risk of venous thromboembolism in older adults. J Gerontol A Biol Sci Med Sci. 2007;62:79–82
- . Complications of preinjury warfarin use in the trauma patient. J Trauma. 2003;54:842–847
- . Advanced age and preinjury warfarin anticoagulation increase the risk of mortality after head trauma. J Trauma. 2006;61:107–110
- . Traumatic brain injury in anticoagulated patients. J Trauma. 2006;60:553–557
- . How I treat venous thromboembolism in patients with cancer. Blood. 2005;106:4027–4033
- . Inferior vena cava filters in cancer patients: indications and outcome. J Clin Oncol. 1996;14:652–657
- . Inferior vena cava filters in malignant disease. J Vasc Surg. 2002;36:704–707
- . Prophylactic indications for vena cava filters: critical appraisal. Semin Vasc Surg. 2005;18:158–165
- . Safety and efficacy of vena cava filters in trauma patients. Injury. 2007;38:7–18
- . Inferior vena caval filter use in US trauma centers: a practitioner survey. J Trauma. 1997;43:333–337
- . Role of prophylactic temporary inferior vena cava filters placed at the ICU bedside under intravascular ultrasound guidance in patients with multiple trauma. J Vasc Surg. 2004;40:958–964
- . Are temporary inferior vena cava filters really temporary?. Am J Surg. 2005;190:858–863
- . Outcomes with retrievable inferior vena cava filters: a multicenter study. J Vasc Interv Radiol. 2006;17:1595–1604
- . Clinical comparison of two optional vena cava filters. J Vasc Interv Radiol. 2007;18:505–511
- . The role of temporary inferior vena cava filters in critically ill surgical patients. Arch Surg. 2003;138:591–594discussion 594-5
Competition of interest: none.
Additional material for this article may be found online at www.jvascsurg.org
PII: S0741-5214(07)01459-0
doi:10.1016/j.jvs.2007.08.057
© 2008 The Society for Vascular Surgery. Published by Elsevier Inc. All rights reserved.
