VTE risk models target a formidable surgery foe

CHICAGO – Implementation of preoperative venous thromboembolism prophylaxis requires a highly individualized approach and a little boost from information technology, one expert suggested.

“What may not be appreciated by some vascular surgeons is that for certain procedures we do, our patients actually have a high VTE [venous thromboembolism] risk. I know people think, ‘We use heparin, so our patients aren’t at high risk,’ but they actually are,” Dr. Melina Kibbe said at a symposium on vascular surgery sponsored by Northwestern University.

The VTE risk is 4.2% for open thoracoabdominal aortic aneurysm repair and 2.2% for thoracic endovascular aortic repair in the American College of Surgeons National Surgical Quality Improvement Program database. That may be an underestimation, however, because the NSQIP database captures only symptomatic events documented by imaging and those events occurring in the first 30 days after surgery.

Patrice Wendling/Frontline Medical News

Smaller studies looking at the problem prospectively suggest the VTE risk is closer to 2%-12% after open aortic surgery and 5.3% after endovascular aortic repair, she said.

Add to that the U.S. Surgeon General’s 2008 call to action to prevent VTE and the Institute of Medicine’s stance that failure to provide VTE prophylaxis qualifies as a medical error, and it would be tempting for vascular surgeons to prescribe pharmacologic prophylaxis for all of their patients, or at least do so based on the type of procedure.

“But VTE formation is really secondary to patient-specific and procedure-related factors,” said Dr. Kibbe, a professor of vascular surgery at Northwestern University in Chicago.

She highlighted five current VTE risk assessment models (RAMs) and the potential returns when RAMs are incorporated into clinical decision support systems.

• Kucher model. One of the earlier and most straightforward RAMs is the Kucher model. It assessed eight weighted risk factors (advanced age, obesity, bed rest, hormone replacement therapy/oral contraceptives, major surgery, cancer, prior VTE, and hypercoagulability) and provided surgeons with electronic alerts regarding prophylaxis.

VTE rates dropped from 8.2% to 4.9% in the high-risk category (score ≥ 4) with the use of the simple physician reminders (N. Engl. J. Med. 2005;352:969-77). Prospective validation showed that VTE increased proportionally with higher scores, Dr. Kibbe said. On the other hand, the model lacked sensitivity at low VTE risk, because 20% of patients with a score of 4 or less actually had VTEs.

“So, while it was simple, it only works well for the high-risk category,” she said.

• Rogers model. One of the most vigorously studied and developed RAMs is the Rogers model (J. Am. Coll. Surg. 2007;204:1211-21). It identified 15 variables (including lab values, patient characteristics, disease states, work relative-value unit, and type of operation) that were independently associated with VTE formation among 183,609 patients undergoing general, vascular, or thoracic procedures at 142 Veterans Health Administration and private hospitals. Each variable is assigned a value from 0 to 9 and added together to create a Rogers score.

Validation showed that VTE risk correlates with the Rogers score, rising from 0.11% for patients with a low score (< 7) to 1.32% for those with a high score (> 10), Dr. Kibbe said. Criticisms of the model are that it is complex, VTE incidence in the entire cohort was low at just 0.63%, the type of VTE prophylaxis used was unclear, and the model lacks prospective validation in a vascular surgery cohort.

• Caprini model. The most commonly used RAM is the 2005 Caprini model (Dis. Mon. 2005;51:70-8), which assigns a weighted score based on more than 30 VTE risk factors compiled by the authors. It has been prospectively validated in numerous studies and shown to accurately stratify 30-day VTE risk at 0.5% for patients at very low risk, 1.5% for low risk, 3% for moderate risk, and 6% for high risk.

The Caprini model, however, was not developed with the same rigor as the Rogers RAM, some of the risk factors have been shown not to be a risk for VTE, and it is complex, Dr. Kibbe observed.

• Pannucci model. The Pannucci model was created specifically to counteract the complexity of the Rogers and Caprini RAMs and incorporates only seven risk factors (personal history of VTE, current cancer, age ≥ 60 years, body mass index ≥ 40 kg/m², male sex, sepsis/septic shock/systemic inflammatory response syndrome, and family history of VTE) into a weighted index for 90-day VTE risk (Chest 2014;145:567-73). The model was developed using a statewide database and a derivation cohort made up of 20% vascular surgery patients.

Both the derivation and validation cohorts identified an 18-fold variation in VTE risk from the lowest- to highest-risk surgical population, showing that the model stratifies patients correctly. Further prospective validation is needed, Dr. Kibbe said.

• Scarborough model. Finally, in an attempt to develop a RAM specific to vascular surgery patients, Dr. John Scarborough and colleagues examined 6,035 patients undergoing open AAA repair in the NSQIP database. The 30-day VTE rate was 2.4% for the entire cohort. Eight independent perioperative risk factors were identified and used to create a nonweighted scoring system (J. Am. Coll. Surg. 2012;214:620-6).

Overall, 65% of patients had 0-1 risk factor and a VTE incidence of 1.5%, while 15% had 3 or more risk factors and a VTE incidence of 6.1%. The Scarborough model has good risk stratification, Dr. Kibbe said, but it is limited by the aforementioned criticisms regarding the NSQIP database, and it also needs prospective validation.

“We all know that proper VTE prophylaxis is very important for our patients; but we need mechanisms by which the attention given to this need for prophylaxis, which is a lot, is turned into proper implementation,” she said.

For Dr. Kibbe and her colleagues, proper implementation meant developing a RAM that was incorporated into the electronic medical record system for all surgical patients at the Jesse Brown VA Medical Center in Chicago. Clinicians were prompted to complete the RAM upon placing orders for preanesthesia testing clearance, and the clinical decision support system would provide a recommended prophylaxis regimen and easily selected electronic orders that could be signed.

A pre- and postimplementation analysis involving 400 consecutive patients revealed an 82% increase in patients with preoperative VTE prophylaxis ordered (22% vs. 40%), a 75% decrease in inappropriate cancellation of orders more than 12 hours before surgery (37% vs. 9%), and a nearly sevenfold increase in the number of patients receiving pharmacologic and mechanical prophylaxis (5% vs. 32%), she said. There was an 80% and 36% decline in DVT rates at 30 and 90 days postoperative, but event rates were too low to detect a significant difference (J. Vasc. Surg. 2010;51:648-54).

Dr. Kibbe reported having no financial disclosures.

pwendling@frontlinemedcom.com

CHICAGO – Implementation of preoperative venous thromboembolism prophylaxis requires a highly individualized approach and a little boost from information technology, one expert suggested.

“What may not be appreciated by some vascular surgeons is that for certain procedures we do, our patients actually have a high VTE [venous thromboembolism] risk. I know people think, ‘We use heparin, so our patients aren’t at high risk,’ but they actually are,” Dr. Melina Kibbe said at a symposium on vascular surgery sponsored by Northwestern University.

The VTE risk is 4.2% for open thoracoabdominal aortic aneurysm repair and 2.2% for thoracic endovascular aortic repair in the American College of Surgeons National Surgical Quality Improvement Program database. That may be an underestimation, however, because the NSQIP database captures only symptomatic events documented by imaging and those events occurring in the first 30 days after surgery.

Patrice Wendling/Frontline Medical News

Smaller studies looking at the problem prospectively suggest the VTE risk is closer to 2%-12% after open aortic surgery and 5.3% after endovascular aortic repair, she said.

Add to that the U.S. Surgeon General’s 2008 call to action to prevent VTE and the Institute of Medicine’s stance that failure to provide VTE prophylaxis qualifies as a medical error, and it would be tempting for vascular surgeons to prescribe pharmacologic prophylaxis for all of their patients, or at least do so based on the type of procedure.

“But VTE formation is really secondary to patient-specific and procedure-related factors,” said Dr. Kibbe, a professor of vascular surgery at Northwestern University in Chicago.

She highlighted five current VTE risk assessment models (RAMs) and the potential returns when RAMs are incorporated into clinical decision support systems.

• Kucher model. One of the earlier and most straightforward RAMs is the Kucher model. It assessed eight weighted risk factors (advanced age, obesity, bed rest, hormone replacement therapy/oral contraceptives, major surgery, cancer, prior VTE, and hypercoagulability) and provided surgeons with electronic alerts regarding prophylaxis.

VTE rates dropped from 8.2% to 4.9% in the high-risk category (score ≥ 4) with the use of the simple physician reminders (N. Engl. J. Med. 2005;352:969-77). Prospective validation showed that VTE increased proportionally with higher scores, Dr. Kibbe said. On the other hand, the model lacked sensitivity at low VTE risk, because 20% of patients with a score of 4 or less actually had VTEs.

“So, while it was simple, it only works well for the high-risk category,” she said.

• Rogers model. One of the most vigorously studied and developed RAMs is the Rogers model (J. Am. Coll. Surg. 2007;204:1211-21). It identified 15 variables (including lab values, patient characteristics, disease states, work relative-value unit, and type of operation) that were independently associated with VTE formation among 183,609 patients undergoing general, vascular, or thoracic procedures at 142 Veterans Health Administration and private hospitals. Each variable is assigned a value from 0 to 9 and added together to create a Rogers score.

Validation showed that VTE risk correlates with the Rogers score, rising from 0.11% for patients with a low score (< 7) to 1.32% for those with a high score (> 10), Dr. Kibbe said. Criticisms of the model are that it is complex, VTE incidence in the entire cohort was low at just 0.63%, the type of VTE prophylaxis used was unclear, and the model lacks prospective validation in a vascular surgery cohort.

• Caprini model. The most commonly used RAM is the 2005 Caprini model (Dis. Mon. 2005;51:70-8), which assigns a weighted score based on more than 30 VTE risk factors compiled by the authors. It has been prospectively validated in numerous studies and shown to accurately stratify 30-day VTE risk at 0.5% for patients at very low risk, 1.5% for low risk, 3% for moderate risk, and 6% for high risk.

The Caprini model, however, was not developed with the same rigor as the Rogers RAM, some of the risk factors have been shown not to be a risk for VTE, and it is complex, Dr. Kibbe observed.

• Pannucci model. The Pannucci model was created specifically to counteract the complexity of the Rogers and Caprini RAMs and incorporates only seven risk factors (personal history of VTE, current cancer, age ≥ 60 years, body mass index ≥ 40 kg/m², male sex, sepsis/septic shock/systemic inflammatory response syndrome, and family history of VTE) into a weighted index for 90-day VTE risk (Chest 2014;145:567-73). The model was developed using a statewide database and a derivation cohort made up of 20% vascular surgery patients.

Both the derivation and validation cohorts identified an 18-fold variation in VTE risk from the lowest- to highest-risk surgical population, showing that the model stratifies patients correctly. Further prospective validation is needed, Dr. Kibbe said.

• Scarborough model. Finally, in an attempt to develop a RAM specific to vascular surgery patients, Dr. John Scarborough and colleagues examined 6,035 patients undergoing open AAA repair in the NSQIP database. The 30-day VTE rate was 2.4% for the entire cohort. Eight independent perioperative risk factors were identified and used to create a nonweighted scoring system (J. Am. Coll. Surg. 2012;214:620-6).

Overall, 65% of patients had 0-1 risk factor and a VTE incidence of 1.5%, while 15% had 3 or more risk factors and a VTE incidence of 6.1%. The Scarborough model has good risk stratification, Dr. Kibbe said, but it is limited by the aforementioned criticisms regarding the NSQIP database, and it also needs prospective validation.

“We all know that proper VTE prophylaxis is very important for our patients; but we need mechanisms by which the attention given to this need for prophylaxis, which is a lot, is turned into proper implementation,” she said.

For Dr. Kibbe and her colleagues, proper implementation meant developing a RAM that was incorporated into the electronic medical record system for all surgical patients at the Jesse Brown VA Medical Center in Chicago. Clinicians were prompted to complete the RAM upon placing orders for preanesthesia testing clearance, and the clinical decision support system would provide a recommended prophylaxis regimen and easily selected electronic orders that could be signed.

A pre- and postimplementation analysis involving 400 consecutive patients revealed an 82% increase in patients with preoperative VTE prophylaxis ordered (22% vs. 40%), a 75% decrease in inappropriate cancellation of orders more than 12 hours before surgery (37% vs. 9%), and a nearly sevenfold increase in the number of patients receiving pharmacologic and mechanical prophylaxis (5% vs. 32%), she said. There was an 80% and 36% decline in DVT rates at 30 and 90 days postoperative, but event rates were too low to detect a significant difference (J. Vasc. Surg. 2010;51:648-54).

Dr. Kibbe reported having no financial disclosures.

pwendling@frontlinemedcom.com

CHICAGO – Implementation of preoperative venous thromboembolism prophylaxis requires a highly individualized approach and a little boost from information technology, one expert suggested.

“What may not be appreciated by some vascular surgeons is that for certain procedures we do, our patients actually have a high VTE [venous thromboembolism] risk. I know people think, ‘We use heparin, so our patients aren’t at high risk,’ but they actually are,” Dr. Melina Kibbe said at a symposium on vascular surgery sponsored by Northwestern University.

The VTE risk is 4.2% for open thoracoabdominal aortic aneurysm repair and 2.2% for thoracic endovascular aortic repair in the American College of Surgeons National Surgical Quality Improvement Program database. That may be an underestimation, however, because the NSQIP database captures only symptomatic events documented by imaging and those events occurring in the first 30 days after surgery.

Patrice Wendling/Frontline Medical News

Smaller studies looking at the problem prospectively suggest the VTE risk is closer to 2%-12% after open aortic surgery and 5.3% after endovascular aortic repair, she said.

Add to that the U.S. Surgeon General’s 2008 call to action to prevent VTE and the Institute of Medicine’s stance that failure to provide VTE prophylaxis qualifies as a medical error, and it would be tempting for vascular surgeons to prescribe pharmacologic prophylaxis for all of their patients, or at least do so based on the type of procedure.

“But VTE formation is really secondary to patient-specific and procedure-related factors,” said Dr. Kibbe, a professor of vascular surgery at Northwestern University in Chicago.

She highlighted five current VTE risk assessment models (RAMs) and the potential returns when RAMs are incorporated into clinical decision support systems.

• Kucher model. One of the earlier and most straightforward RAMs is the Kucher model. It assessed eight weighted risk factors (advanced age, obesity, bed rest, hormone replacement therapy/oral contraceptives, major surgery, cancer, prior VTE, and hypercoagulability) and provided surgeons with electronic alerts regarding prophylaxis.

VTE rates dropped from 8.2% to 4.9% in the high-risk category (score ≥ 4) with the use of the simple physician reminders (N. Engl. J. Med. 2005;352:969-77). Prospective validation showed that VTE increased proportionally with higher scores, Dr. Kibbe said. On the other hand, the model lacked sensitivity at low VTE risk, because 20% of patients with a score of 4 or less actually had VTEs.

“So, while it was simple, it only works well for the high-risk category,” she said.

• Rogers model. One of the most vigorously studied and developed RAMs is the Rogers model (J. Am. Coll. Surg. 2007;204:1211-21). It identified 15 variables (including lab values, patient characteristics, disease states, work relative-value unit, and type of operation) that were independently associated with VTE formation among 183,609 patients undergoing general, vascular, or thoracic procedures at 142 Veterans Health Administration and private hospitals. Each variable is assigned a value from 0 to 9 and added together to create a Rogers score.

Validation showed that VTE risk correlates with the Rogers score, rising from 0.11% for patients with a low score (< 7) to 1.32% for those with a high score (> 10), Dr. Kibbe said. Criticisms of the model are that it is complex, VTE incidence in the entire cohort was low at just 0.63%, the type of VTE prophylaxis used was unclear, and the model lacks prospective validation in a vascular surgery cohort.

• Caprini model. The most commonly used RAM is the 2005 Caprini model (Dis. Mon. 2005;51:70-8), which assigns a weighted score based on more than 30 VTE risk factors compiled by the authors. It has been prospectively validated in numerous studies and shown to accurately stratify 30-day VTE risk at 0.5% for patients at very low risk, 1.5% for low risk, 3% for moderate risk, and 6% for high risk.

The Caprini model, however, was not developed with the same rigor as the Rogers RAM, some of the risk factors have been shown not to be a risk for VTE, and it is complex, Dr. Kibbe observed.

• Pannucci model. The Pannucci model was created specifically to counteract the complexity of the Rogers and Caprini RAMs and incorporates only seven risk factors (personal history of VTE, current cancer, age ≥ 60 years, body mass index ≥ 40 kg/m², male sex, sepsis/septic shock/systemic inflammatory response syndrome, and family history of VTE) into a weighted index for 90-day VTE risk (Chest 2014;145:567-73). The model was developed using a statewide database and a derivation cohort made up of 20% vascular surgery patients.

Both the derivation and validation cohorts identified an 18-fold variation in VTE risk from the lowest- to highest-risk surgical population, showing that the model stratifies patients correctly. Further prospective validation is needed, Dr. Kibbe said.

• Scarborough model. Finally, in an attempt to develop a RAM specific to vascular surgery patients, Dr. John Scarborough and colleagues examined 6,035 patients undergoing open AAA repair in the NSQIP database. The 30-day VTE rate was 2.4% for the entire cohort. Eight independent perioperative risk factors were identified and used to create a nonweighted scoring system (J. Am. Coll. Surg. 2012;214:620-6).

Overall, 65% of patients had 0-1 risk factor and a VTE incidence of 1.5%, while 15% had 3 or more risk factors and a VTE incidence of 6.1%. The Scarborough model has good risk stratification, Dr. Kibbe said, but it is limited by the aforementioned criticisms regarding the NSQIP database, and it also needs prospective validation.

“We all know that proper VTE prophylaxis is very important for our patients; but we need mechanisms by which the attention given to this need for prophylaxis, which is a lot, is turned into proper implementation,” she said.

For Dr. Kibbe and her colleagues, proper implementation meant developing a RAM that was incorporated into the electronic medical record system for all surgical patients at the Jesse Brown VA Medical Center in Chicago. Clinicians were prompted to complete the RAM upon placing orders for preanesthesia testing clearance, and the clinical decision support system would provide a recommended prophylaxis regimen and easily selected electronic orders that could be signed.

A pre- and postimplementation analysis involving 400 consecutive patients revealed an 82% increase in patients with preoperative VTE prophylaxis ordered (22% vs. 40%), a 75% decrease in inappropriate cancellation of orders more than 12 hours before surgery (37% vs. 9%), and a nearly sevenfold increase in the number of patients receiving pharmacologic and mechanical prophylaxis (5% vs. 32%), she said. There was an 80% and 36% decline in DVT rates at 30 and 90 days postoperative, but event rates were too low to detect a significant difference (J. Vasc. Surg. 2010;51:648-54).

Dr. Kibbe reported having no financial disclosures.

pwendling@frontlinemedcom.com