Extended indications for placement of an inferior vena cava filter Michael J. R o h r e r , M D , Michael G. Scheidler, M D , H . BrowneU Wheeler, M D , and Bruce S. Cutler, M D , Worcester, Mass. To study the morbidity and mortality rates after placement of an inferior vena cava filter and to define the appropriate indications for interruption of the inferior vena cava, the records of all patients who underwent insertion of a Greenfield filter during the decade January 1978 to December 1987 were reviewed. Patients were designated as having either a traditional or extended indication for placement of an inferior vena cava filter. Two hundred sixty inferior vena cava filters were placed in 264 attempts, with no deaths related to insertion of the filter. An extended indication was the primary reason for placement of the Greenfield filter in 66 (25%) of the patients. In patients with extended indications there were no cases o f air embolism or filter misplacement and only three wound complications (4.5%). Pulmonary embolism after insertion of the inferior vena cava filter occurred in three patients (4.5%), with one fatality (1.5%). Inferior vena cava occlusion was documented in three cases (4.5%), and manifestations of the postphlebitic syndrome in early follow-up were present in two patients (3.0%). As the procedures to prevent fatal pulmonary embolism have become safer, more efficacious, and less morbid, the number of patients in whom the potential benefits of insertion of an inferior vena cava filter outweigh the risks has become larger. Our results support the liberalized use of Greenfield filters in those patients who do not necessarily have one o f the traditional indications for placement of an inferior vena cava filter but are at a high risk of having a fatal pulmonary embolus. (J VASC SURG 1989;10:44-50.)
Indications for interruption o f the inferior vena cava (WC) have traditionally included the presence o f a contraindication to anticoagulation in a patient who has proximal deep venous thrombosis (DVT) or pulmonary embolus (PE), or recurrent PE while being therapeutically anticoagulated. Interruption o f the IVC has also been advocated after pulmonary embolectomy or whenever a large freefloating iliofemoral thrombus is noted on a venogram. 18 It is well recognized that the presence o f any o f these indications clearly identifies a patient who is at high risk for a fatal PE. 7"9 The ease and safety o f transvenous placement o f a Greenfield filter have encouraged us to consider liberalizing the indications for interruption o f the IVC to include patients who lacked traditional indications for the procedure but could still be considered to be at risk for pulmonary embolism. The purFrom the Division of Vascular Surgery, Universityof Massachusetts Medical School. Presented at the Fifteenth Annual Meeting of the New England Societyfor VascularSurgery,Montreal, Quebec, Canada, Sept. 15-16, 1988. Reprint requests: Michael J. Rohrer, MD, Division of Vascular Surgery, Universityof Massachusetts Medical Center, 55 Lake Ave. N., Worcester,MA 01655. 44
pose o f the study is to evaluate the potential benefit for patients with extended indications for interruption o f the W C and to study the morbidity, mortality, and cost-effectiveness o f the procedure. MATERIAL AND METHODS The records o f all patients who underwent placement o f a Greenfield filter at the University o f Massachusetts Medical Center for the decade January 1978 to December 1987 were reviewed. The indications for placement o f a Greenfield filter, intraoperative complications, clinical course, and followup morbidity and mortality relating to placement o f a Greenfield filter were specifically noted. Patients were assigned to one o f two groups based on the indications for placement o f a Greenfield filter. Group I consisted o f patients in whom the primary indication for placement o f a Greenfield filter was a conventional indication for interruption o f the IVC. Group II was composed o f patients for whom the primary indication for placement o f a Greenfield filter was not one o f the traditional reasons for interruption o f the IVC. Group II included (1) patients with no documented D V T or PE who were at high risk and who could not receive D V T prophylaxis or surveillance; (2) patients at risk for D V T / P E in w h o m
Volume 10 Number 1 July 1989
even a small PE would likely be fatal because of their very low cardiopulmonary reserve, as defined by the presence of a cardiac index of less than 1.5 L/min, pulmonary compromise requiring ventilatory support, or the administration of steroids for chronic obstructive pulmonary disease; (3) patients with documented DVT but no PE who could be anticoagulated but were at high risk for life-threatening PE because of ifiofemoral thrombus location or morphology; (4) patients having the procedure in conjunction with venous thrombectomy; and (5) patients with thrombus identified above a previously placed IVC filter (Table I). The presence of DVT was routinely documented by means of noninvasive methods including impedance plethysmography and venous duplex scanning. Venography was performed in equivocal cases. Pulmonary emboli were identified by nuclear ventilation/perfusion scanning showing a high probability of PE or by pulmonary arteriography. Comparisons were made between the cases from group I and group II for frequency of insertion and perioperative morbidity and mortality. The rates of late morbidity including caval patency and chronic venous stasis sequelae were also compared. Although IVC patency was not routinely documented, when symptoms of thrombosis such as recurrent PE or leg swelling were noted, or symptoms of the postphlebitic syndrome were observed, IVC patency was assessed by venography. Postphlebitic changes were classified as severe if venous ulceration or swelling refractory to use of a support stocking occurred. Statistical comparisons were made with the Fisher exact test. RESULTS
From 1978 to 1987 260 Greenfield filters were successfully placed in 264 attempts. The right internal jugular approach was used for 195 of the filters. Sixty-two Greenfield filters were placed from the right common femoral vein, nine of which followed a failed internal jugular approach. The right atrium was used to introduce the filter carrier in three cases when IVC filters were placed in patients undergoing cardiac operations. Placement of a Greenfield filter was unsuccessful in four patients because of the inability to pass the filter cartier into the IVC. A second IVC filter was placed in four patients because ofmalposition of the first Greenfield filter and in two others because of recurrent PE with thrombus proximal to the original fiker. The number of IVC filters placed for both traditional and[ extended indications increased substantially over the study period (Fig. 1). A traditional
Placement of an inferior vena cava filter
Table I. Extended indications for placement of IVC filters in 66 patients Cases Indication Undocumented DVT or PE, but at high risk No prophylaxis or surveillance possible Low cardiopulmonary reserve Documented DVT, no contraindication to anticoagulation At high risk for PE After venous thrombectomy Thrombus above previously placed IVC filter Low cardiopulrnonary reserve Documented PE, no contraindication to anticoagulants Pulmonary hypertension as a result of chronic pulmonary embolic disease Low cardiopulmonary reserve Total
9 2 2
14% 3% 3%
indication was the primary reason for the placement of a Greenfield filter in 198 patients (75%). An extended indication was the primary reason for placement of a filter in the other 66 cases (25%). Of the 198 cases in which a traditional indication for placement of a Greenfield filter was present, 170 (86%) involved ,either a relative or absolute contraindication to anticoagulation in the setting of a documented DVT or PE. Only 14 patients had a recurrent PE while they were therapeutically anticoagulated as their primary indication for placement of an IVC filter. Six patients had placement of a Greenfield filter after pulmonary embolectomy, and eight had a large free-floating iliofemoral thrombus identified on a venogram. Twenty-one of the 66 patients having IVC filters implanted for extended indications were at risk for DVT and PE because of extensive pelvic and lower extremity fractures and could not receive subcutaneous heparin prophylaxis for DVT because of associated head trauma. Orthopedic hardware frequently prevented the use of intermittent compression boots ,or even routine impedance plethysmography or wmous duplex surveillance for DVT. Others at risk tbr PE included two patients undergoing venous thrombectomy for phlegmasia cerulea dolens and two patients with venographically proven IVC thrombus above a previously placed IVC filter who had recurrent PE documented by artetiography. Nine patients with extensive iliofemoral DVT were thought to be at high risk for PE and had IVC filters placed. Other extended indications for placement of an
Jourrlal of VASCULAR SURGERY
46 Rohrer et al.
Year Fig. 1. The number of !VC filters placed annually generally increased from 1978 to 1987, both for traditional and extended indications.
IVC filter were present in 32 patients who were at increased risk of death from even a small PE because of a low cardiopulmonary reserve. This group included 16 patients with previous massive acute PE, eight patients with chronic recurrent pulmonary emboli and pulmonary hypertension, and eight others with severe chronic obstructive pup monary disease. No perioperative deaths occurred among patients with either traditional or extended indications attributable to placement of a Greenfield filter. However, two deaths occurred in group I within 24 hours of placement of a Greenfield filter in patients who had undergone pulmonary embolectomy. The perioperative morbidity associated with placement of a filter in group I patients included four episodes of air embolism, three instances of superficial wound infection, five cases of wound hematoma, two cases of unilateral vocal cord paralysis, and four cases of malposition of the filter. Three of the four misplaced filters were placed in the right renal vein, and the fourth was discharged in the right common iliac vein. Each of these cases was managed by placement of a second IVC filter in the correct position. Morbidity in group II patients was limited to three cases of wound hematoma. There were no episodes of arrhythmia requiring treatment in either group (Table II). Late morbidity included six cases of pulmonary embolism after placement of a filter in group I (3.0%), with one fatality, and three cases of pulmonary embolism in group II (4.5%) (p = 0.23), with one fatality, for an overall failure rate of 3.4%. In the
two fatal cases the PE was documented at the time of autopsy. Nonfatal recurrent pulmonary emboli were identified by either arteriography or a change in the ventilation/perfusion scan, which suggested a high probability of PE when compared to previous studies. Ten patients in group I had documented IVC occlusion in follow-up (5.0%), and three caval occlusions occurred in group II patients (4.5%), for an overall IVC patency rate of 95 %. Phlegmasia cerulea dolens occurred three times but predated placement of the filter in each case, and was therefore not considered to be a complication of placement of the Greenfield filter. Migration of the filter or embolization was not noted. Late development of postphlebitic symptoms of a moderate or severe degree (refractory swelling or venous stasis ulceration) was noted in the follow-up of six group I patients (3.0%) and two group II patients (3.0%) (Table III). DISCUSSION
Anticoagulation is still the mainstay of therapy for DVT and PE, and in most patients it is very effective at preventing the fatal complications of thromboembolic disease, The occurrence of pulmonary embolism while therapeutic anticoagulation is being carried out has been documented in 2% to 12% of all patients with iliofemoral thrombosis, 1'912 with a fatal pulmonary embolism rate of less than 2%.1,5,9-14
However, those patients who cannot be anticoagulated because of concurrent medical problems or who have had a failure of anticoagulant therapy are at risk for a fatal PE. The most dramatic examples
Volume 10 Number 1 July 1989
Placement of an inferior vena cava filter
Table III. Late complications of placement of IVC filter in 264 patients
Table II. Operative complications of placement of IVC filter in 264 patients
Symptomatic air embolus Filter malposition Unilateral vocal cord paralysis Wound hematoma Wound infection Total
Group I (n = 198)
Group H (n = 66)
4 4 2 5 3 18
0 0 0 3 (4.5%) 0 3 (4.5%)
(2.0%) (2.0%) (l.0%) (2,5%) (1.5%) (9.0%)
are those clinical conditions that define the traditional indications for interruption of the IVC. Patients with iliofemoral DVT or PE who cannot be anticoagulated have approximately a 50% chance of having a PE and an 18% to 26% risk of death from pulmonary embolism.9,14 The mortality in those patients who have a PE in spite of therapeutic coagulation is ~increased dramatically with each recurrencc,7 and the likclihood of further recurrence increases with each subsequent event.7'15 Norris et al.s report that patients with free-floating iliofemoral DVT have a 60% chance of having a symptomatic PE within 10 days of making the diagnosis in spite of initiating anticoagulant therapy. Furthermore, in a small series Greenfield et al.16 reported a 25% mortality from recurrent PE in patients undergoing pulmonary embolectomy. These very high-risk clinical conditions prompted the consideration of mechanical interruption of the IVC. Before the development of transvenous methods of caval interruption, a transabdominal or rctroperitoneal surgical caval ligation or plication was the only effective alternative to anticoagulation to prevent fatal puhnonary embolization. These procedures carried a substantial mortality, commensurate with the performance of a major operation in patients who were frequently both acutely and chronically ill. Thc mortality rates average 9.9% and 14.3% for caval plication and caval ligation, respectively. 7'17 I n a& dition, venous stasis sequelae after caval ligation or plication were very common and led to refractory leg swelling and the postphlebitic syndrome in as many as 45% of the patients. ~7'~8Because of the morbidity and mortality of surgical caval interruption, the indications for caval interruption were strictly limited to those patients who were at extremely high risk for fatal pulmonary embolism, namely, those patients with the traditional indications for interruption of the IVC. A recognition of the risk of postoperative DVT and PE has prompted several authors to advocate routine placement of an IVC clip or plication at the
Group I (n = 198)
Symptomatic IVC occlusion Recurrent nonfatal PE Recurrent fatal PE Postphlebiticsyndrome Total
10 5 1 6 22
(5.0%) (2.5%) (0.5%) (3.0%) (11.0%)
Group I I (n = 66)
3 2 1 2 9
(4.5%) (3.0%) (1.5%) (3.0%) (12.0%)
time of laparotomy in patients who do not have one of the traditional indications for interruption of the IVC but who do have multiple risk factors for the development of DVT and PE, 19-21 including patients undergoing pelvic operations for malignancy,22 colonic surgery, 23 and even routine aortic reconstruction? 4,2~Although elective placement of an IVC clip at the time of another operation adds little in tcrms of operative morbidity and mortality to the primary operative procedure,18-2sthe high ratc of IVC thrombosis and venous stasis sequelae limits the usefulness of this approach. 17'18 The development and evolution of transvenous mcans of caval interruption has substantially improved both the morbidity and mortality of the procedure. Mortality from insertion of a Greenfield filter is less than 1%, and long-term caval patency is at least 95%. 24,26 Despite this improved safcty, there have been few reports to date that suggest the liberalization of the indications for interruption of the IVC. We have identified specific groups of patients who are at risk for DVT and PE who do not have one of the traditional indications for interruption of the IVC but who would benefit from placement of an IVC filter. For examplc, the designation of our medical center as a level I regional trauma center in 1981 resulted in a markcd increase in the number of admissions for blunt trauma. The nature of many of the injuries, which frequently included pelvic or extremity fracturcs in combination with abdominal or neurologic injury, often precluded the use of conventional DVT prophylaxis or even surveillance. Although these paticnts did not have traditional indications for interruption of the IVC, previous studies have shown that the risk of developing DVT approaches 100% when the radioactive fibrinogcn technique is used27'28and ranges from 20% to 90% when venography is used. 283~ The risk of pulmonary embolization in these patients has bcen reported to range from 4% to 22%, 28m-33with a fatal pulmonary embolism rate ranging from 2% to as high as 16%.27'33 As a conscquencc, our indications for place-
Rohrer et al.
ment of an IVC filter were extended to indude this group of patients. Assuming a fatal pulmonary embolism incidence of 5%, this group of 66 patients with extended indications for placement of an IVC filter might have been expected to have had as many as three or four episodes of fatal pulmonary embolism if unprotected by a Greenfield filter. In our series only one patient in this high-risk group had a fatal pulmonary embolism with the IVC filter in place. Patients with extensive iliofemoral D V T or with thrombus above a previously placed IVC filter were fek to be at a sufficiently high risk for pulmonary embolism to justify placement of an IVC filter. Similarly, patients who underwent venous thrombectomy were fek to have an extended indication for placement of a Greenfield filter. Each of these conditions is remarkable for thrombus in a proximal location of sufficient size to lead to a fatal PE. These impressions are supported by the work of Alexander et al. 6 who concluded that patients who have had an initial PE with D V T of sufficient magnitude to be identifiable by phleborheography and Doppler venous studies should have an IVC filter placed because of the high risk of recurrent PE, in spite of the ability to provide therapeutic anticoagulation. Indications for placement of an IVC filter were also extended to include patients with severe pulmonary hypertension and cor pulmonale in whom the ability to tolerate even a small PE is doubtful. TM Assuming a risk of PE of 2% to 12% in patients with iliofemoral D V T even with therapeutic anticoagulation, 1,912 placement of an IVC filter is justified in this group of patients. Prophylactic placement of a transvenous IVC filter has also been advocated for patients in specific high-risk situations such as hip and pelvic fractures 27~28and for patients with iliofemoral thrombosis and cancer. 34 Golueke et al.36 have recently advocated a more liberalized use of the Greenfield filter based on their observations of effectiveness and safety in 88 cases of insertion of the filter. They too have noted no deaths related to placement of a transvenous IVC filter and an operative complication rate of only 4.6%, which is almost identical to the 4.5% in the present series of 66 cases with extended indications. The incidence of recurrent pulmonary embolism, caval occlusion, and venous stasis disease was also similar. Inferior vena cava filters were implanted for extended indications in 23.9% of the patients of Golueke et al. compared to 25% in our own series. The cost of our aggressive approach to PE prophylaxis is high. Total patient charges for placement
Journal of VASCULAR SURGERY
of an IVC filter are approximately $4000, which includes $1350 for the filter and insertion hardware, $2000 for the surgeon's and anesthesiologist's professional fees, and $650 for hospital operating room charges. If one assumes that the rate o f fatal PE may be as high as 5%, it would be necessary to place 20 IVC filters at a cost of approximately $80,000 to prevent one death because of PE, if the Greenfield filter were 100% effective at preventing a fatal PE. This high cost forces one to examine the cost-effectiveness of this liberalized attitude toward placement of an IVC filter. In a young otherwise healthy patient suffering trauma the additional $4000 expense may be only a small fraction of the overall cost of hospitalization, and the $80,000 cost per fatal PE prevented may be considered a reasonable expense. On the other hand, one must critically examine the indications for placement of an IVC closely when placing this expensive device in a patient who has a limited life expectancy, such as a terminally ill patient with cancer. In the latter setting the IVC filter may only prevent PE as the cause of death in a small percentage of the patients who will go on to die of their underlying illnesses in any event. Based on a favorable long-term experience with transvenous caval interruption with the Greenfield filter, we believe that it is reasonable to extend traditional indications for placement of a Greenfield filter to these patients at risk for D V T and PE with those extended indications for placement of an IVC filter outlined in Table I. There is evidence from our own experience and others that a Greenfield filter can be inserted with an operative mortality rate approaching zero and a fatal failure rate of 1.5%. Therefore we feel that it is reasonable to place a Greenfield filter in any patient with an otherwise good life expectancy whose risk of death from PE is 3% or greater. We recognize that there is the potential for abuse in liberalizing indications for insertion of a Greenfield filter, particularly with respect to increased costs of health care. However, we also feel that it is unacceptable to allow patients with an otherwise good prognosis to die because of failure to use a simple prophylactic procedure with very low morbidity and no mortality. REFERENCES 1. Greenfield LJ, Alexander EL. Current status of surgical therapy for deep vein thrombosis. Am J Surg 1985;150:64-70. 2. Greenfield LJ. Current indications for and results of Greenfield filter placement. J Vasc St;rtc 1984;1:502-4. 3. Steward JR, Peyton JWR, Crute SL, Greenfield LJ. Clinical results of suprarenal placement of the Greenfield vena cava filter. Surgery 1982;92:1-4,
Volume 10 Number 1 July 1989
4. Greenfield LI, Peyton R, Crute S, Barnes R. Greenfield vena caval filter experience: late results in 156 patients. Arch Surg 1981;116:1451-6. 5. Jones RH, Sabiston DC Jr. Pulmonary embolism. Surg Clin North Am 1976;56:891-907. 6. Alexander JJ, Gewertz BL, Lu CT, Zarins CK. New criteria for placement of a prophylactic vena cava filter. Surg Gynecol Obstet 1986;163:405-9. 7. Jones TK, Barnes RW, Greenfield LJ. Greenfield vena cava filter: rationale and current indications. Ann Thorac Surg 1986;42:$48-55. 8. Norris CS, Greenfield LJ, Herrmann JB. Free-floating ileofemoral thrombus: a risk of pulmonary embolism. Arch Surg i985;120:806-8. 9. Einarsson E, Eklof B. Acute ileofemoral venous thrombosis. In: Eiseman B, ed. Prognosis of surgical disease. Philadelphia: WB Saunders, 1980:117-9. 10. Adar R, Salzmon EW. Treatment of thrombosis of veins of lower extremities. N Engl I Med 1975;292:348-50. 11. Kelly G. Postoperative pulmonary embolus. In: Eiseman B, ed. Prognosis of surgical disease. Philadelphia: WB Saunders, 1980:113-6. 12. Coon WN, Willis III PW, Symons MJ. Assessment of anticoagulant treatment of venous thromboembolism. Ann Surg 1969;170:559-68. 13. Kernohan RJ, Todd C. Heparin therapy in thromboembolic disease. Lancet 1966;1:621-3. 14. Barritt DW, Jordan SC. Anticoagulant drugs in the treatment of pulmonary embolism: a controlled trial. Lancet 1960; 1:1309-12. 15. Coon WWI Willis HI PW. Recurrence of venous thromboembolism. Smgery 1973;73:823-7. I6. Greenfield LJ, Peyton MD, Brown PP, Elkins RC. Transvenous management of pulmonary embolic disease. Ann Surg 1974;180:461-8. 17. DeWeese M& Complications of caval ligation and transabdominal partial interruption of the vena cava. In: Bernard VM, Towne JB, eds. Complications in vascular surgery. New York: Grune & Stratton, Inc, 1980:311-24. 18. Carmichael JD, Edwards WS. Prophylactic inferior vena caval plication. Surg Gynecol Obstet 1967;124:785-6. 19. Rao G, Zikria EA, Miller WH, Samadani SR, Ford WB. Concomitant prophylactic inferior vena cava clipping. Int Surg 1976;61:147-50. 20. McDowell HA. Prophylactic Teflon clip plication of the inferior vena cava for prevention of pulmonary embolism. South Med J 1973;66:356-8.
DISCUSSION Dr. James O. M e n z o i a n (Boston, Mass.). The excellent results with placement o f Greenfield fiker in these 260 patients are commendable, but I think could pose a significant problem. If we are n o t careful, we may have a new indication for interruption o f the I V C and that may be that it is just so simple and so safe and so easy. I do n o t mean to exaggerate, b u t in our institution for every cava we interrupt we turn d o w n one request to do another one. It is an extremely popular procedure. Internists, family
Placement of an inferior vena cava filter 4 9
21. Fuller CH, Willbanks OL. Incidental prophylactic inferior vena cava clipping. Arch Surg 1971;102:440-1. 22. Lindenauer SM. Prophylactic staple plication of the inferior vena cava. Arch Surg 1973;107:669-75. 23. Kusminsky RE, Medina S, AbuRahma AF, Boland lV. Prophylactic inferior vena cava clipping in colonic surgery. Dis Colon Rectum 1982;25:108-12. 24. Korwin SM, CaUow AD, Rosenthal D, Ledig CB, Deterling RA Jr, O'Donnell TF Jr. Prophylactic interruption of the inferior vena cava: immediate and long term hcmodynamic effects. Arch Surg 1979;114:1037-40. 25. Rosenthal D, Cossman D, Matsumoto G, Callow AD. Prophylactic interruption of the inferior vena cava: a retrospective evaluation. Am J Surg 1979;137:389-93. 26. Stewart JR, Greenfield LJ. Transvenous vena caval filtration and pulmonary embolectomy. Surg Clin North Am 1982; 62:411-30. 27. Todd JW, Frisbie JH, Rossier AB, et al. Deep venous thrombosis in acute spinal cord injury: a comparison of 1125 fibrinogen leg scanning impedance plethysmography and venography. Paraplegia 1976;14:50-7. 28. Myllynen P, Kalnmonen M, Rokkanen P, Bostman O, Lalla M, Laasonen E. Deep venous thrombosis in patients with acute spinal cord injury: a comparison with non-paralyzed patients immobili2~ed due to spinal fractures. J Trauma 1985;25:541-3. 29. Brach BB, Moser YAM,Cedar L, Minteer M, Covery R. Venous thrombosis in acute spinal cord paralysis. J Trauma
1977;17:289-92. 30. Williams JW. Venous thrombosis and pulmonary embolism. Surg Gynecol Obstet 1975;14i:626-30. 31. Freeark RJ, Boswick J, Fardin R. Post-traumatic venous thrombosis. Arch Surg 1967;95:567-75. 32. Shackford SR, Moser KM. Deep venous thrombosis and pulmonary embolus ha trauma patients. J Intensive Care Med 1988;3:87-98. 33. Hyers TM, Hull l~J), Weg IG. Antithrombotic therapy for venous thrombocmbolic disease. Chest 1989;95(suppl):37S49S. 34. FuUen WD, Miller EH, Steele WF, McDonough JJ. Prophylactic vena caval interruption in hip fractures. J Trauma 1973;13:403-10. 35. Davis FE. Pelvic fractures at a community hospital. Am Surg 1986;52: 548-50. 36. Golueke PJ, Garrer~ WV, Thompson IE, Smith BL, Talkinton CM. Interruption of the vena cava by means of the Greenfield fiker: expanding the indications. Surgery 1988; 103:111-7.
practitioners, everybody loves it because it seems to be a small innocuous procedure and once it is done no one has to worry about DVT, or pulmonary emboli. But I would like to suggest that although it is extremely safe, with essentially a zero moixality rate, the morbidity rate is n o t zero. There are some patients in this series and our o w n series with w o u n d infection, with air emboli, and with occlusion o f the vena cava. There are extended indications for interruption o f the vena cava. For example, in patients with malignancies and established DVT, it may be difficult
Journal of VASCULAR SURGERY
Rohrer et aL
to administer long-term chemotherapy and sodium warfarin (Coumadin). Therefore we have recommended caval interruption for those patients. In addition, there are also extended or soft indications in patients with an established PE in whom it is felt that because of age or instabiliw, long-term Coumadin is probably hazardous. If a patient has a very massive PE and is hemodynamically unstable, I think all of us would feel very strongly about caval interruption. These may not really be extended indications. Chronic recurrent pulmonary emboli are also a reasonable indication for caval interruption. I have some concerns about the inclusion of patients with limited cardiopulmonary reserve or chronic obstructive lung disease. Are you suggesting that all patients with bilateral lower extremity fractures who cannot undergo routine surveillance and cannot undergo compression should undergo routine interruption of the vena cava? Likewise, do you feel that patients with head trauma and spinal cord injury should undergo routine interruption of the cava? We believe that this group o f patients present an ideal situation for surveillance by impedance plethysmography (IPG), phleborheography; or duplex scanning. Are your patients with iliofemoral DVT at a higher risk for pulmonary emboli? Data we presented to this Society a few years ago indicated that there was not a higher incidence of subsequent pulmonary emboli in those patients with proximal DVT. H o w do you select those patients with chronic obstructive pulmonary disease who should undergo prophylactic routine caval interruption? And finally, does your institution have any experience with the percutaneous method of caval interruption? I think that there are unquestionably extended indications, but it is up to us to really define these indications. I think this paper is a step in the right direction. Dr. Michael Hurtle (Boston, Mass.). Under the diagnosis-related group system, early discharge after major surgery is affecting the incidence of venous thrombosis and in some instances pulmonary embolism. We are now seeing the readmission of patients after an a~hroscopy or some other major procedure in which they return to the hospital with a thrombosis or an embolism. The initial postoperative and postdischarge care needs to be planned with the same foresight that the in-hospital care has been accorded over the years. Dr. H. Brownell Wheeler (Worcester, Mass.). I would like to ask if anyone else in the audience has had experience with placement of a percutaneous filter? Dr. Menzoian's cautions about things becoming too quick, simple, and reassuring and therefore too widely applied are apt to be doubly true when placement of the filter becomes a percutaneous procedure. I know for example that there have been aggressive radiologists putting in Greenfield filter through a 24F catheter in the femoral vein, and I think the indications are getting dangerously thin. I have been surprised there have not been more reported complications fi'om such procedures, but I have heard anecdotally about one center where noninvasive tests were performed on patients who have had these percutaneous catheters placed through a large sheath. A high incidence
of occlusion at the site of placement was found, even though the patients often had no symptoms. The new titanium Greenfield filter with a 12F introducer is going to be very hard to keep away from the catheterization laboratories and the fluoroscopy suites, and the indications may become much too liberal.
Dr. Joseph M. Van De Water (Philadelphia, Pa.). There is now a word of caution coming from Duke University. At Duke they do not encourage the use of vena caval filtration at all. They had put in one filter in the entire year of 1987 in their hospital. They feel that the major problem that patients have is pulmonary hypertension occurring from thrombosis in the pulmonary arteries or even very small emboli that would not be stopped by a filter.
Dr. H, Brownell Wheeler. At Duke an extraordinarily low incidence of pulmonary emboli has been reported. We are completing a retrospective review of 16 hospitals in our geographic region, and pulmonary embolism does remain a major cause of death in these hospitals. Certainly Duke is an outstanding medical center, but their experience does not conform to other centers in this regard. Dr. John Bergan (Lalolla, Calif.). Percutaneous placement is very dangerous. Fifty percent of people suffer t?om a new venous thrombosis at the site, proved by noninvasive detection, and 50% of these have symptoms. At Northwestern we saw an air embolus and a pelvic abscess and a new pulmonary embolization after percutaneous placement. I want you all to be aware that if it is being done in your hospital there is a lot of danger. Dr. Michael J. Rohrer (closing). To put all this into perspective, we still believe that heparin is a very effective means of treating DVT. With therapeutic anticoagulation we expect the incidence of PE to be only 2%. It is only in special patients that we consider placement of an IVC filter. We do not think all patients with bilateral lower extremity fractures should have W C filters, but those patients with bilateral extremity fractures who cannot receive any sort of DVT prophylaxis or surveillance probably should. The use of duplex scanning to image the popliteal, femoral, and iliac segments in those patients with tibial and fibular fractures has enabled us to provide some form of DVT surveillance other than IPG testing, and it has allowed us to avoid placing an IVC filter in some patients. As far as spinal cord injury is concerned, we all recognize that those patients are at extremely high risk for developing DVT. The incidence may be as high as 75%, but those patients too are routinely followed with surveillance and only under special circumstances are filters placed. I understand that there are spinal cord centers where an IVC filter is part of the treatment protocol. Finally, iliofemoral thrombosis is not treated routinely with W C filters at our institution. However, Dr. Zarins at the University of Chicago has advocated that patients with a positive IPG and the diagnosis of pulmonary embolus should have an IVC filter placed because he feels that they are at high risk for recurrent PE in spite of the fact that they can be therapeutically anticoagulated.