Inferior vena cava filters: a framework for evidence-based use

Despite advances in prevention strategies, venous thromboembolism (VTE) remains a leading cause of preventable hospital mortality.¹  In most patients, pharmacologic and/or mechanical thromboprophylaxis are sufficient to prevent VTE. For acute VTE, anticoagulation remains the treatment of choice without need for additional interventions. However, in patients with active bleeding or high risk of bleeding, inferior vena cava filters (IVCFs) are used to mechanically interrupt the inferior vena cava (IVC), thereby preventing pulmonary embolism (PE). Unlike anticoagulation, IVCFs neither treat VTE nor prevent deep vein thrombosis (DVT) or in situ PE.

Designing an IVCF that can be deployed safely, traps thrombi, preserves laminar flow, and minimizes the inherent thrombogenicity of an intravascular device is challenging. IVC interruption advanced over the course of 100 years from surgical IVC ligation to percutaneous placement of permanent filters (eg, Greenfield filter). In 2003, the first retrievable inferior vena cava filter (rIVCF) was approved for patients at risk for VTE with short-term contraindication to anticoagulation. rIVCFs have largely supplanted permanent inferior vena cava filters (pIVCFs), although there is no evidence that they are either safer or more effective.¹  IVCF use has increased due to liberalized indications, bedside placement techniques, increased numbers of specialists with skills of insertion, improved detection of PE with modern imaging, and the unconfirmed belief that rIVCFs are safer than older pIVCFs. Therapeutic indications for IVCFs have increased linearly, whereas placement for prophylactic indications has increased supralinearly.²  Globally, the United States surpassed the 5 largest European nations in IVCF insertion by 25-fold despite similar annual VTE mortality.²  Herein, we present safety and efficacy data surrounding IVCFs in common clinical scenarios, complications of IVCFs, considerations for timely IVCF retrieval, and a holistic approach for how to incorporate IVCFs into practice (Figure 1; Table 1).

IVCF placement can be grouped into 2 categories: (1) therapeutic indications for known VTE and (2) prophylactic indications for patients labeled “high risk” for VTE. Because of a dearth of high-quality studies, recommendations from clinical practice guidelines are incongruent, leading to wide practice variation (Table 2).^3-15 

A 63-year-old-man presents 2 days after a fall with headache, vision changes, and nausea. Computed tomography of his head reveals an acute, moderate-sized right subdural hematoma (SDH) without midline shift. Two days after admission, he complains of right leg tenderness and is diagnosed with an acute right iliofemoral vein DVT. The result of computed tomography with pulmonary angiography is negative for pulmonary embolism. An IVCF is considered.

The standard of care in patients with acute VTE is therapeutic anticoagulation.⁸  However, when anticoagulation is contraindicated, such as in major bleeding or emergent surgery, anticoagulation may be delayed or interrupted. Estimated rates of recurrent VTE without anticoagulation are 40% in the first month and 10% in the second and third month after the diagnosis of acute VTE.¹⁶  Large observational studies evaluating the benefit of IVCFs in patients with acute VTE and contraindications to anticoagulation have reported conflicting results, and no prospective studies have been performed.^17,18  Nevertheless, as in case 1, therapeutic IVCF placement for acute VTE with contraindication to anticoagulation is the only consensus indication for routine IVCF placement.^4-9,11,13-15  In some cases in which therapeutic anticoagulation is temporarily contraindicated, prophylactic dose anticoagulation with titration to a therapeutic dose may be preferable to IVCF placement.

Data on the use of IVCFs in patients with known VTE come predominantly from 2 randomized controlled trials (RCTs) that explored IVCFs as an adjunct to therapeutic anticoagulation (Table 3).^19-21  In PREPIC, patients randomized to a pIVCF compared with no filter had a sustained reduction in PE at the cost of long-term increase in DVT and no change in mortality. This benefit would likely be diminished with currently recommended durations of anticoagulation for similar high-risk patients.^19,20  In PREPIC 2, the low rate of recurrent PE observed in the nonfilter group is consistent with successful contemporary therapeutic anticoagulation.²¹  Taken together, the results of PREPIC and PREPIC 2 do not provide justification for routine IVCF placement for VTE that can be treated with anticoagulation.

Malignancy is an independent risk factor for VTE. The increasing incidence of VTE is possibly due to longer survival of patients with cancer, administration of prothrombotic systemic therapies, and improved VTE diagnostic measures. The high frequency of recurrent VTE and bleeding cannot be explained by over- or under-anticoagulation.¹³  Low-molecular-weight heparins or direct oral anticoagulants are the preferred anticoagulants for cancer-associated thrombosis.²²  A meta-analysis found no difference in recurrent VTE in patients with cancer receiving an IVCF as an adjunct to anticoagulation.²³  A small RCT explored the benefits of an IVCF in addition to fondaparinux compared with fondaparinux alone in patients with cancer with acute DVT and reported higher DVT resolution rates in the nonfilter arm without any difference in PE, DVT, or 90-day mortality.²⁴  Furthermore, because the hypercoagulable state of cancer affects all vascular beds, regional therapies, such as IVCFs, are likely insufficient for prevention of recurrent thrombosis and may instead be thrombogenic. Nevertheless, patients with cancer are twice as likely as patients without cancer to receive an IVCF, and retrieval rates are lower.²⁵  Evidence-based guidelines recommend against the routine use of IVCFs in cancer-associated thrombosis outside of classic indications.^9,13 

Recurrent VTE while receiving therapeutic anticoagulation is uncommon and often represents inadequate treatment or noncompliance. Patients with suspected recurrent VTE should continue therapeutic anticoagulation without IVCF placement, ensuring attention to proper dosing, malabsorption issues, medication adherence, and potential drug or food interactions that may reduce anticoagulant efficacy. Current and past imaging should be compared to distinguish acute from chronic thrombosis. In confirmed anticoagulation failure, it is preferable to use alternative anticoagulants or dose escalation.⁸  In such cases, hypercoagulable conditions, such as active malignancy, antiphospholipid syndrome, heparin-induced thrombocytopenia, paroxysmal nocturnal hemoglobinuria, or myeloproliferative syndromes, should be explored.²⁶  In prothrombic conditions, IVCFs will not prevent recurrent VTE and may serve as a thrombotic stimulus. Data from the PREPIC and PREPIC2 provide ample evidence that supplementing anticoagulation with IVCFs did not improve outcomes.^19-21  Recurrent lower-extremity DVT or non–catheter-related upper-extremity DVT in the same location, especially in young patients, should prompt evaluation for vascular anomalies such as May-Thurner syndrome (abnormal compression of the left common iliac vein by the right iliac artery) or Paget-Schroetter syndrome (thoracic outlet syndrome). In these anatomic variants, thrombolysis, pharmacomechanical thrombectomy, endovascular stenting, and decompressive surgery are preferred interventions over IVCFs, which can exacerbate the underlying chronic obstruction to venous flow.²⁶ 

IVCFs have been proposed for major surgery requiring interruption of anticoagulation. Clinicians must determine if the surgery is urgent or elective and the necessary duration off anticoagulation. If surgery can be delayed to allow completion of 3 months of therapeutic anticoagulation or if anticoagulation will be held for only a short period of time (eg, <48 hours for a low–bleeding risk procedure), aggressive pharmacologic prophylaxis with dose escalation to therapeutic doses may be preferred over an IVCF. If the surgery cannot be deferred in the setting of recently diagnosed DVT (≤30 days), an rIVCF can be considered.^4,10  The rationale for this approach is based on the estimated 40% risk of VTE recurrence in the first month after diagnosis.¹⁶  If an rIVCF is placed, a systematic plan should be delineated before insertion to ensure timely removal and avoid filters being left indwelling permanently. Beyond 30 days, VTE recurrence risk is reduced, and thus the risks of IVCF at that point likely outweigh the benefits.

Patients with massive PE (large-volume PE accompanied by hemodynamic instability) are considered at high risk for fatal recurrent PE and thus are potential candidates for rIVCFs. The concern is that additional PE could lead to poor outcomes due to limited cardiopulmonary reserve. However, most patients with massive PEs will receive emergent systemic or catheter-directed thrombolysis (CDT) and promptly experience reduced thrombus burden, questioning the utility of an IVCF at that point. Large registry studies have provided mixed results but overall have demonstrated short-term survival benefit in patients with massive PE, regardless of thrombolytic use.²⁷  However, given the inherent selection and survival biases of registry studies, we cannot currently recommend routine use of rIVCFs as short-term adjuncts to anticoagulation and thrombolysis.^5,8,15,27 

IVCF deployment before CDT has been suggested as an appropriate indication. Retrospective studies of CDT combined with IVCFs showed no difference in PE or mortality compared with CDT alone, but increased complications were noted in patients who had filters placed.^28,29  Conversely, the FILTER-PEVI RCT demonstrated that IVCFs lowered the incidence of immediate post-procedural symptomatic PE without mortality benefit compared with patients receiving anticoagulation alone.³⁰  The authors advised a selective approach to filter placement in those with specific predictors of PE. IVCFs have also been considered for proximal free-floating DVT; however, anticoagulation alone is sufficient for treatment.^4,6 

IVCF placement for primary VTE prophylaxis is controversial but accounts for more than half of IVCFs placed in the United States.³¹  The rationale for inserting a prophylactic IVCF is to offer mechanical protection against PE during the limited high-risk period when pharmacologic prophylaxis may be contraindicated.

A 42-year-old morbidly obese woman (body mass index, 55 kg/m²) presents for elective gastric bypass surgery. She has no personal or family history of VTE. The surgeon asks whether a rIVCF preoperatively for VTE prophylaxis is appropriate.

VTE is an important cause of preventable postoperative mortality after bariatric surgery owing to obesity, immobility, surgery, and possible underdosing with standard pharmacologic prophylaxis. The reported incidence of DVT is 1% to 3% and that of PE is 0.3% to 2%, but mortality with PE may be as high as 40%.³²  Incidence of PE is highest within 1 month after bariatric surgery and may occur despite pharmacologic prophylaxis. A practice pattern survey revealed that 28% of respondents used IVCFs routinely before bariatric surgery.³³  No RCTs of prophylactic IVCFs in this population exist, leading to incongruent guideline recommendations.^7,12,15  A National Inpatient Sample study and a meta-analysis reported no difference in PE, increased rates of DVT, and increased risk of mortality, calling into question empiric placement of IVCFs before bariatric surgery.^32,34  Therefore, in case 2, aggressive pharmacologic prophylaxis rather than an IVCF should be used for prevention of postoperative VTE.

A similar lack of benefit of prophylactic placement of IVCFs is apparent in other high-risk surgical patients. In major trauma, VTE occurs in up to 58% of patients without thromboprophylaxis.³⁵  Although pharmacologic prophylaxis is effective and unanimously recommended by evidence-based guidelines, many patients with trauma have ongoing or perceived risk of bleeding and are not considered to be candidates for initial anticoagulation, owing to their underlying injuries.^3,7,12  Conflicting guidelines on IVCF use in patients with trauma has led to inconsistent practice patterns.^3,7,10-12  Three meta-analyses and 2 RCTs, did not demonstrate a reduction in fatal PE or death with prophylactic IVCF placement in patients with trauma (Table 3).^36-40  In recent years, IVCF use in patients with trauma has declined without an increase in PE rates, further supporting a restrictive strategy.⁴¹  For additional information on IVCF use in trauma, refer to the evidence-based minireview in Hematology 2020.⁴² 

Prophylactic IVCFs also have unproven benefit perioperatively in patients undergoing spinal surgery, total hip arthroplasty, or total knee arthroplasty and are therefore not recommended.^7,11,12,15  Importantly, IVCF insertion may lead to a delay in initiation of pharmacologic prophylaxis.

A 39-year-old man presents with abdominal pain and melena. He has a history of rIVCF placement 7 years ago for primary VTE prevention after a motor vehicle accident. The IVCF was never removed. Computed tomography of his abdomen and pelvis reveals IVCF tines extending beyond the wall of the IVC into the lumen of the duodenum. Esophagogastroduodenoscopy confirms that the IVCF perforated the distal duodenum with evidence of recent bleeding. General surgery is consulted for laparotomy with IVCF removal.

Recognizing the paucity of evidence showing benefit of IVCFs in most circumstances, providers must consider the mounting evidence for adverse events with these devices. Complications may occur in the immediate post-procedural period, early after IVCF placement, or years later (Table 4). Immediate and early complications are uncommon, and fatal complications are rare, occurring in only 0.12% of insertions.⁴³  Late complications are more common, particularly when filters are left indwelling beyond when risk–benefit analysis favors removal. Recurrent DVT, even in patients who are receiving anticoagulation, may reflect filter-mediated changes in venous flow, the underlying hypercoagulable condition of the patient, or a synergistic effect of both. Furthermore, PE can still occur despite the presence of an IVCF due to thrombus extension proximally off the device.

From 2009 to 2012, 1606 IVCF-related adverse events were reported to the US Food and Drug Administration Manufacturer and User Facility Device Experience database on IVCF complications.⁴⁴  These rates likely underestimate the true incidence of complication rates because reporting is voluntary. Published rates of specific IVCF complications are disparate due to variance in filter types, follow-up duration, complication definitions, use of concurrent anticoagulation, and use of screening imaging. Thrombotic and device-related adverse events are 6 times more likely to be reported with indwelling rIVCFs than with pIVCFs (86.8% vs 13.2%; P < .0001).⁴⁴  Optimal management of nonthrombotic device-related complications is unknown. Management decisions should be made in collaboration with the interventionalist on a case-by-case basis, weighing risks of intervention vs continued monitoring in asymptomatic patients.

The patient in case 1 with acute SDH and acute lower-extremity DVT has an rIVCF placed. He recovers gradually from SDH without operative intervention. Two weeks later, neurosurgery is comfortable with initiating anticoagulation. The patient remains stable after initiation of therapeutic anticoagulation without any signs of recurrent bleeding.

Over 50% of IVCFs are placed for temporary prevention of VTE, but only 12% to 45% are retrieved.⁴⁵  This reflects a combination of overconfidence in the long-term safety of indwelling rIVCFs, lack of provider and patient education on the importance of retrieving filters, and loss to follow-up.^31,45  When attempted, >90% of IVCF retrievals are successful in the first month. At 12 months, the success rate drops to 37%.³¹  Procedural factors associated with retrieval failure include prolonged dwelling time, advanced patient age, filter tilting, adherence to the IVC wall, or large clot volume within the filter.³¹  Clinical factors that influence the rate of IVCF retrieval include comorbidities, concurrent anticoagulation, insurance coverage, primary indication for placement, and documented plans for removal.⁴⁵  The urgency of early retrieval was highlighted in a 2014 US Food and Drug Administration safety alert.⁴⁶  Though no guidelines recommend a specific time frame for removal, a decision analysis study found that retrieval between 29 and 54 days after insertion was optimal.⁴⁷  Notably, an IVCF left indwelling permanently is not of itself an indication for indefinite anticoagulation.^4,7  The underlying thrombotic event and perceived bleeding risk should guide duration of anticoagulation.

To improve provider- and system-related factors associated with low retrieval rates, providers have focused on increased clinician education and oversight, novel technical aspects of retrieval, and streamlining systems-based approaches for patient follow-up.^44,48  Poor patient education is a barrier to IVCF removal. In one qualitative study 12% of patients interviewed were not aware of having an IVCF, 77% did not know an IVCF can be removed, and 79% were not aware of long-term risks of IVCFs, highlighting patient education and engagement as an important strategy to improve retrieval rates.⁴⁹  Ultimately, the decision to retrieve an IVCF should be based on the patient’s current risk of thrombosis and bleeding risk with anticoagulation. In case 1b, the patient has initiated anticoagulation for DVT without progression of SDH; therefore, IVCF retrieval should be planned. Optimizing appropriate IVCF removal requires a collaborative approach with multidisciplinary providers and shared decision making with the patient.

Despite the availability of safe and effective anticoagulants, a small group of patients with acute VTE and absolute contraindication to anticoagulation will continue to require IVCFs. However, for extended indications, there is insufficient evidence to support routine IVCF use. If IVCFs are employed, close follow-up is vital with attention to resumption of anticoagulation when safe, monitoring for filter complications, and IVCF removal when no longer needed. Further research is essential to address these popular, but overall unsubstantiated, devices (Table 5). Until well-designed trials are available, IVCFs will remain a contentious topic. Clearly, a “more is better” approach is not appropriate when incorporating IVCFs into clinical practice, and therefore clinicians must assimilate the available evidence to make case-by-case decisions.

Anita Rajasekhar, University of Florida – Shands Hospital, 1600 SW Archer Rd, Gainesville, FL 32610; e-mail: rajasa@medicine.ufl.edu.