DOI: 10.3171/2014.6.JNS131419 ©AANS, 2014
What clinical factors predict the incidence of deep venous thrombosis and pulmonary embolism in neurosurgical patients? Clinical article John D. Rolston, M.D., Ph.D.,1 Seunggu J. Han, M.D.,1 Orin Bloch, M.D., 2 and Andrew T. Parsa, M.D., Ph.D. 2 Department of Neurological Surgery, University of California, San Francisco, California; and 2Department of Neurological Surgery, Northwestern University, Chicago, Illinois
1
Object. Venous thromboembolisms (VTEs) occur frequently in surgical patients and can manifest as pulmonary emboli (PEs) or deep venous thromboses (DVTs). While many medical therapies have been shown to prevent VTEs, neurosurgeons are concerned about the use of anticoagulants in the postoperative setting. To better understand the prevalence of and the patient-level risk factors for VTE, the authors analyzed data from the National Surgical Quality Improvement Program (NSQIP). Methods. Retrospective data on 1,777,035 patients for the years from 2006 to 2011 were acquired from the American College of Surgeons NSQIP database. Neurosurgical cases were extracted by querying the data for which the surgical specialty was listed as “neurological surgery.” Univariate statistics were calculated using the chi-square test, with 95% confidence intervals used for the resultant risk ratios. Multivariate models were constructed using binary logistic regression with a maximum number of 20 iterations. Results. Venous thromboembolisms were found in 1.7% of neurosurgical patients, with DVTs roughly twice as common as PEs (1.3% vs 0.6%, respectively). Significant independent predictors included ventilator dependence, immobility (that is, quadriparesis, hemiparesis, or paraparesis), chronic steroid use, and sepsis. The risk of VTE was significantly higher in patients who had undergone cranial procedures (3.4%) than in those who had undergone spinal procedures (1.1%). Conclusions. Venous thromboembolism is a common complication in neurosurgical patients, and the frequency has not changed appreciably over the past several years. Many factors were identified as independently predictive of VTEs in this population: ventilator dependence, immobility, and malignancy. Less anticipated predictors included chronic steroid use and sepsis. Venous thromboembolisms appear significantly more likely to occur in patients undergoing cranial procedures than in those undergoing spinal procedures. A better appreciation of the prevalence of and the risk factors for VTEs in neurosurgical patients will allow targeting of interventions and a better understanding of which patients are most at risk. (http://thejns.org/doi/abs/10.3171/2014.6.JNS131419)
Key Words • pulmonary embolus • deep venous thrombosis • venous thromboembolism • complication • quality improvement • vascular disorders
N
eurosurgical patients are at risk for developing potentially catastrophic deep venous thromboses (DVTs) and pulmonary emboli (PEs).7,13 Unfractionated heparin, low-molecular-weight heparin, compression stockings, and sequential compression devices all re-
Abbreviations used in this paper: ACS = American College of Surgeons; COPD = chronic obstructive pulmonary disease; DVT = deep venous thrombosis; NSQIP = National Surgical Quality Improvement Program; PE = pulmonary embolus; SIRS = systemic inflammatory response syndrome; VTE = venous thromboembolism.
J Neurosurg / August 1, 2014
duce the risk of DVTs and PEs.2,4,7,13 Some of these prophylactic treatments bear specific risks, including intracerebral hemorrhage and bleeding, and concerns about these risks can limit the use of these treatments.2,6–8,13 Here we used data from the American College of Surgeons (ACS) National Surgical Quality Improvement Program (NSQIP)9,10 to identify predictors of developing venous thromboembolism (VTE) and analyzed VTE occurrence over time. Efficient management paradigms will probably depend on identifying those patients most at risk for developing VTE. The long-term goal of the present study is to facilitate a better understanding of which patients will most likely 1
J. D. Rolston et al. benefit from prophylactic treatments that mean added costs to health care while potentially incurring risk.
Methods
Data on 1,777,035 patients for the years 2006 to 2011 were acquired from the ACS NSQIP database.9,10 This database contains a nationwide data set representing cases from a wide range of academic and private hospitals across 47 US states.14 Cases are analyzed for a predetermined collection of complications, including VTEs. Reviewers are frequently audited, and stringent criteria for each complication are used. In particular, DVTs must be diagnosed within 30 days of an operation by using duplex ultrasound, venography, or CT scanning, and patients must be treated for these clots with an inferior vena cava filter, inferior vena cava ligation, or anticoagulation therapy. Pulmonary embolisms must occur within 30 days of an operation and must be confirmed with a highprobability ventilation-perfusion (VQ) scan, CT scan, or pulmonary arteriogram. Database files were acquired in delimited text format and parsed using both SPSS version 20 (IBM Corp.) and MATLAB R2012a (MathWorks Inc.). Neurosurgical cases were extracted by querying the data for which the surgical specialty was listed as “neurological surgery” and further classified as “spine” or “cranial” depending on the current procedural terminology (CPT) code (Table 1). Descriptive statistics are represented as the means ± standard deviation. Univariate statistics were calculated using the chi-square test, and 95% confidence intervals are presented for the resultant risk ratios. Predictors included in the models are listed in Table 2. Multivariate models were constructed using binary logistic regression with a maximum of 20 iterations. Variables that were significant (p < 0.05) on univariate analysis were included in the multivariate model. During model building, insignificant predictors (p > 0.05) were removed in a stepwise fashion using the Wald statistic. Significant predictors from this model were identified, and their odds ratios were presented along with the 95% confidence intervals.
Results Demographics and Prevalence
Using the NSQIP database, we identified 38,058 neurosurgical cases for the years 2006 to 2011. Patient demographics are shown in Table 3. Six hundred forty cases of VTE were identified, corresponding to 1.7% of the neurosurgical cases. Of these cases, 244 (0.6%) involved PEs and 484 (1.3%) involved DVTs requiring treatment. Asymptomatic and untreated DVTs were not included in this analysis. The prevalence of VTEs was consistent from year to year, ranging from a low of 1.3% in 2007 to a high of 1.8% in 2008. There was no significant trend in the percentage of VTEs over time (F = 0.759; p = 0.45). This consistency in prevalence was also true for DVTs (min 1.0% in 2007 and max 1.5% in 2008; F = 0.305; p = 0.62) and PEs (min 0.4% in 2007 and max 0.7% in 2009; F = 0.304; p = 0.62).
2
TABLE 1: Current procedural terminology (CPT) codes included in study Spine Codes
Cranial Codes
0090T 0092T 0093T 0095T 0096T 0163T 0195T 0196T 0202T 10121 10140 10180 11000 11042-4 20005 20100 20696 20902 20922 20930 20931 20936 20938 20999 21026 21501 21510 21610 21615 22010 22015 22100-14 22206-26 22318-27 22520-95 22600-32 22800-99 27080 27202 27218 27280 42725 49215 63001-308 63685 63700-41 64573-85
21137 21139 21180 21299 35001-5 35201 35301 35800 37605 37799 49422 60600 61150-888 62000-362 69511
J Neurosurg / August 1, 2014
Pulmonary embolism and deep venous thrombosis in neurosurgery TABLE 2: Predictors included in models* Patient Characteristic sex race type of surgery age year of operation height weight diabetes tobacco use ethanol use dyspnea ventilator dependence in 48 hrs prior to surgery severe COPD preop pneumonia ascites esophageal varices congestive heart failure recent MI prior PCI prior cardiac surgery angina hypertension requiring medication peripheral vascular disease rest pain renal failure dialysis altered mental status coma hemiparesis TIA stroke w/ persistent deficit stroke w/o persistent deficit CNS tumor paraparesis quadriparesis disseminated cancer wound infection chronic steroid use recent weight loss bleeding disorder transfusion chemotherapy radiotherapy sepsis
Value M, F white, African American, Asian or Pacific Islander, American Indian or Alaskan Native cranial, spinal 16–90 yrs 2006–2011
none, insulin independent, insulin dependent no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes none, SIRS, sepsis, septic shock
* MI = myocardial infarction; PCI = percutaneous coronary intervention; TIA = transient ischemic attack.
J Neurosurg / August 1, 2014
TABLE 3: Basic demographics for 38,058 neurosurgical patients in the NSQIP database* Characteristic
No. (%)
mean age in yrs ± SD mean weight in lbs ± SD mean height in inches ± SD sex M F unknown race white African American Asian or Pacific Islander American Indian or Alaskan Native unknown surgery type cranial spinal unknown inpatient vs outpatient inpatient outpatient unknown diabetes no yes insulin dependent not insulin dependent unknown tobacco use no yes unknown ethanol use no yes unknown ventilator dependence in 48 hrs before op no yes unknown severe COPD no yes unknown preop pneumonia no yes unknown
56.0 ± 15.0 186.7 ± 46.9 66.9 ± 4.2 19,349 (50.8) 18,634 (49.0) 75 (0.2) 30,104 (79.1) 2922 (7.7) 722 (1.9) 221 (0.6) 4089 (10.7) 10,041 (26.4) 28,017 (73.6) 0 32,628 (85.7) 5430 (14.3) 0 32,630 (85.7) 5428 (14.3) 1906 (5.0) 3522 (9.3) 0 28,434 (74.7) 9623 (25.3) 1 (0.0) 26,530 (69.7) 1064 (2.8) 10,464 (27.5) 37,401 (98.3) 657 (1.7) 0 36,507 (95.9) 1551 (4.1) 0 27,505 (72.3) 69 (0.2) 10,484 (27.5) (continued)
3
J. D. Rolston et al. TABLE 3: Basic demographics for 38,058 neurosurgical patients in the NSQIP database* (continued) Characteristic ascites no yes unknown esophageal varices no yes unknown CHF no yes unknown prior MI no yes unknown prior PCI no yes unknown prior cardiac surgery no yes unknown angina no yes unknown hypertension requiring medication no yes unknown peripheral vascular disease no yes unknown rest pain no yes unknown renal failure no yes unknown dialysis no yes unknown
No. (%) 38,037 (99.9) 21 (0.1) 0 27,566 (72.4) 8 (0.0) 10,484 (27.5) 37,931 (99.7) 127 (0.3) 0 27,497 (72.3) 77 (0.2) 10,484 (27.5) 26,173 (68.8) 1401 (3.7) 10,484 (27.5) 26,450 (69.5) 1123 (3.0) 10,485 (27.6) 27,430 (72.1) 143 (0.4) 10,485 (27.6) 20,205 (53.1) 17,853 (46.9) 0 27,311 (71.8) 263 (0.7) 10,484 (27.5) 27,521 (72.3) 52 (0.1) 10,485 (27.6) 38,003 (99.9) 54 (0.1) 1 (0.0) 37,899 (99.6) 158 (0.4) 1 (0.0) (continued)
4
TABLE 3: Basic demographics for 38,058 neurosurgical patients in the NSQIP database* (continued) Characteristic
No. (%)
altered mental status no yes unknown coma no yes unknown hemiparesis before surgery no yes unknown prior TIA no yes unknown history of stroke w/ persistent deficit no yes unknown history of stroke w/o persistent deficit no yes unknown CNS tumor no yes unknown paraparesis no yes unknown quadriparesis no yes unknown disseminated cancer no yes unknown wound infection no yes unknown steroid use no yes unknown
26,524 (69.7) 1050 (2.8) 10,484 (27.5) 27,350 (71.9) 224 (0.6) 10,484 (27.5) 26,396 (69.4) 1178 (3.1) 10,484 (27.5) 26,918 (70.7) 656 (1.7) 10,484 (27.5) 26,441 (69.5) 1133 (3.0) 10,484 (27.5) 27,048 (71.1) 526 (1.4) 10,484 (27.5) 24,669 (64.8) 2905 (7.6) 10,484 (27.5) 26,463 (69.5) 1111 (2.9) 10,484 (27.5) 27,339 (71.8) 235 (0.6) 10,484 (27.5) 36,527 (96.0) 1531 (4.0) 0 37,427 (98.3) 631 (1.7) 0 35,740 (93.9) 2318 (6.1) 0 (continued)
J Neurosurg / August 1, 2014
Pulmonary embolism and deep venous thrombosis in neurosurgery TABLE 3: Basic demographics for 38,058 neurosurgical patients in the NSQIP database* (continued) Characteristic recent weight loss no yes unknown bleeding disorder no yes unknown transfusion no yes unknown chemotherapy no yes unknown radiotherapy no yes unknown sepsis none SIRS sepsis septic shock unknown emergency case no yes
No. (%) 37,633 (98.9) 425 (1.1) 0 36,966 (97.1) 1091 (2.9) 1 (0.0) 37,913 (99.6) 145 (0.4) 0 27,300 (71.7) 274 (0.7) 10,484 (27.5) 27,218 (71.5) 173 (0.5) 10,667 (28.0 36,581 (96.1) 896 (2.4) 281 (0.7) 59 (0.2) 241 (0.6) 35,806 (94.1) 2252 (5.9)
* CHF = congestive heart failure.
Predictors of VTE
Multivariate statistical analysis was used to identify patient characteristics predictive of VTE, PE, and DVT formation. For VTEs (that is, DVTs requiring treatment and PEs), the rate of cranial cases was far higher than the rate of spinal cases (3.4% vs 1.1%; Table 4). This difference was significant, with an OR of 0.44 (95% CI 0.34– 0.57), showing the lower likelihood of a VTE developing in spinal patients as compared with cranial patients. This difference persisted when analyzing solely DVTs or PEs. For DVTs alone, there were events in 259 cranial cases (2.6%) versus 225 spinal cases (0.8%; OR 0.52, 95% CI 0.39–0.69; Table 5). For PEs alone, there were events in 137 cranial cases (1.4%) as compared with 107 spinal cases (0.4%; OR 0.34, 95% CI 0.22–0.51; Table 6). Other significant predictors of VTE included the presence of a CNS tumor (OR 2.24, 95% CI 1.71–2.94), any significant motor weakness (whether hemiparesis [OR 1.80, 95% CI 1.32–2.45], quadriparesis [OR 3.78,
J Neurosurg / August 1, 2014
95% CI 2.00–7.17] or paraparesis [OR 2.56, 95% CI 1.78– 3.69]), chronic steroid use (OR 1.63, 95% CI 1.20–2.21), and ventilator dependence prior to surgery (OR 2.41, 95% CI 1.60–3.62; Table 4). Other notable predictors were the presence of systemic inflammatory response syndrome (SIRS) criteria (OR 2.32, 95% CI 1.61–3.32), sepsis (OR 3.97, 95% CI 2.25–7.02), or septic shock (OR 3.58, 95% CI 1.32–9.81). These individual predictors were also identified when constructing a multivariate model exclusively with DVTs (Table 5). However, three additional significant protective predictors were found: severe chronic obstructive pulmonary disease (COPD; OR 1.62, 95% CI 1.05–2.49), chemotherapy use (OR 2.16, 95% CI 1.20–3.88), and tobacco use (OR 0.70, 95% CI 0.51–0.96). Because there were fewer PEs than DVTs (244 vs 484) and consequently less power in the statistical analysis, fewer significant predictors were identified in the analysis of PEs when examined in isolation (Table 6). Nevertheless, the type of procedure (cranial vs spinal) remained significant, as did the presence of paraparesis (OR 2.78, 95% CI 1.57–4.90) or SIRS (OR 2.69, 95% CI 1.56–4.63). One additional predictor, not found in the analysis of either VTEs or DVTs, was an altered mental status (OR 1.93, 95% CI 1.17–3.18). Again, these predictors were identified even when confounding variables, such as the presence of disseminated cancer or CNS tumors, were controlled for.
Discussion
Using the NSQIP database, which contains data from more than 1 million surgical patients, we analyzed patient-level predictors of VTE and, individually, of DVT and PE. Patients who underwent spine surgery were significantly less likely to develop DVTs or PEs than those who underwent cranial surgery, with an OR of 0.44 (95% CI 0.34–0.57). This was true even when accounting for multiple confounders using multivariate logistic regression (see Table 2 for a list of variables). The cause of this difference is unknown. Perhaps the NSQIP data set is lacking important predictors of VTE that would otherwise account for the discrepancy, although a great deal of variables are included, such as the presence of a CNS tumor, paralysis, stroke, coma, ventilator dependence, and so forth (Table 3). Alternatively, physician practice could explain some of the difference. Perhaps cranial surgeons are less likely than spinal surgeons to prescribe VTE prophylaxis. If so, this tendency could be attributable to a heightened concern among cranial surgeons for postoperative bleeding and the devastating consequences thereof, despite the fact that multiple studies refute this idea.1,3,5,7,12 It should be noted that the NSQIP database contains no information about whether thromboprophylaxis was used. A last possibility relates to follow-up time. Perhaps cranial patients are hospitalized longer and thus have more time for VTEs to present themselves in the hospital setting and to show up as complications within the medical record. Other predictors of VTE identified in this study have been previously mentioned and thus offer validity to the methodology used herein. These predictors include the presence of cancer and venostasis, such as that found in 5
J. D. Rolston et al. TABLE 4: Significant predictors of VTE in neurosurgical patients, both univariate and multivariate Factor age weight surgery type cranial spinal hemiparesis before surgery no yes history of stroke w/o persistent deficit no yes chemotherapy no yes ventilator dependence in 48 hrs before op no yes CNS tumor no yes paraparesis no yes quadriparesis no yes steroid use no yes sepsis none SIRS sepsis septic shock severe COPD no yes tobacco use no yes patient sex M F race white African American Asian or Pacific Islander American Indian or Alaskan Native
No. of VTEs (%)
Univariate OR (95% CI)
Multivariate OR (95% CI)† 1.03 (1.02–1.03)* 1.01 (1.01–1.01)*
345 (3.4) 295 (1.1)
1 [reference] 0.30 (0.26–0.35)*
1 [reference] 0.44 (0.34–0.57)*
387 (1.5) 70 (5.9)
1 [reference] 4.25 (3.27–5.52)*
1 [reference] 1.80 (1.32–2.45)*
437 (1.6) 20 (3.8)
1 [reference] 2.41 (1.52–3.80)*
1 [reference] 1.85 (1.12–3.05)*
437 (1.6) 20 (7.3)
1 [reference] 4.84 (3.04–7.71)*
1 [reference] 1.94 (1.12–3.36)*
581 (1.6) 59 (9.0)
1 [reference] 6.25 (4.73–8.27)*
1 [reference] 2.41 (1.60–3.62)*
322 (1.3) 135 (4.6)
1 [reference] 3.69 (3.00–4.52)*
1 [reference] 2.24 (1.71–2.94)*
412 (1.6) 45 (4.1)
1 [reference] 2.67 (1.95–3.65)*
1 [reference] 2.56 (1.78–3.69)*
445 (1.6) 12 (5.1)
1 [reference] 3.25 (1.81–5.86)*
1 [reference] 3.78 (2.00–7.17)*
550 (1.5) 90 (3.9)
1 [reference] 2.58 (2.06–3.24)*
1 [reference] 1.63 (1.20–2.21)*
538 (1.5) 64 (7.1) 24 (8.5) 6 (10.2)
1 [reference] 5.15 (3.94–6.74)* 6.26 (4.08–9.59)* 7.58 (3.25–17.72)*
1 [reference] 2.32 (1.61–3.32)* 3.97 (2.25–7.02)* 3.58 (1.32–9.81)*
592 (1.6) 48 (3.1)
1 [reference] 1.94 (1.44–2.61)*
532 (1.9) 108 (1.1)
1 [reference] 0.60 (0.48–0.73)*
353 (1.8) 286 (1.5)
1 [reference] 0.84 (0.72–0.98)*
487 (1.6) 81 (2.8) 8 (1.1) 3 (1.4)
1 [reference] 1.74 (1.37–2.20)* 0.68 (0.34–1.38) 0.84 (0.27–2.62) (continued)
6
J Neurosurg / August 1, 2014
Pulmonary embolism and deep venous thrombosis in neurosurgery TABLE 4: Significant predictors of VTE in neurosurgical patients, both univariate and multivariate (continued) Factor ethanol use no yes hypertension requiring medication no yes preop pneumonia no yes dialysis no yes renal failure no yes altered mental status no yes coma no yes history of TIA no yes history of stroke w/ persistent deficit no yes disseminated cancer no yes wound infection no yes recent weight loss no yes bleeding disorder no yes transfusion no yes radiotherapy no yes CHF no yes
No. of VTEs (%)
Univariate OR (95% CI)
443 (1.7) 14 (1.3)
1 [reference] 0.79 (0.46–1.34)
279 (1.4) 361 (2.0)
1 [reference] 1.47 (1.26–1.73)*
453 (1.7) 4 (5.8)
1 [reference] 3.67 (1.33–10.13)*
638 (1.7) 2 (1.3)
1 [reference] 0.75 (0.19–3.03)
637 (1.7) 3 (5.6)
1 [reference] 3.45 (1.07–11.08)*
383 (1.4) 74 (7.0
1 [reference] 5.17 (4.00–6.69)*
438 (1.6) 19 (8.5)
1 [reference] 5.69 (3.53–9.20)*
438 (1.6) 19 (2.9)
1 [reference] 1.80 (1.13–2.87)*
394 (1.5) 63 (5.6)
1 [reference] 3.89 (2.96–5.11)*
575 (1.6) 65 (4.2)
1 [reference] 2.77 (2.13–3.60)*
616 (1.6) 24 (3.8)
1 [reference] 2.36 (1.56–3.58)*
629 (1.7) 11 (2.6)
1 [reference] 1.56 (0.85–2.86)
592 (1.6) 48 (4.4)
1 [reference] 2.83 (2.09–3.82)*
632 (1.7) 8 (5.5)
1 [reference] 3.44 (1.68–7.06)*
441 (1.6) 12 (6.9)
1 [reference] 4.53 (2.50–8.20)*
635 (1.7) 5 (3.9)
1 [reference] 2.41 (0.98–5.91)
Multivariate OR (95% CI)†
(continued)
J Neurosurg / August 1, 2014
7
J. D. Rolston et al. TABLE 4: Significant predictors of VTE in neurosurgical patients, both univariate and multivariate (continued) Factor
No. of VTEs (%)
Univariate OR (95% CI)
455 (1.7) 2 (2.6)
1 [reference] 1.58 (0.39–6.47)
436 (1.7) 21 (1.5)
1 [reference] 0.90 (0.58–1.40)
425 (1.6) 31 (2.8)
1 [reference] 1.74 (1.20–2.56)*
452 (1.6) 4 (2.8)
1 [reference] 1.72 (0.63–4.67)
451 (1.7) 6 (2.3)
1 [reference] 1.39 (0.62–3.14)
455 (1.7) 1 (1.9)
1 [reference] 1.17 (0.16–8.46)
456 (1.7) 1 (12.5)
1 [reference] 8.49 (1.04–69.17)*
prior MI no yes prior PCI no yes prior cardiac surgery no yes angina no yes peripheral vascular disease no yes rest pain no yes esophageal varices no yes ascites no yes
639 (1.7) 1 (4.8)
Multivariate OR (95% CI)†
1 [reference] 2.93 (0.39–21.84)
* Indicates a statistically significant odds ratio. † Constant for multivariable model was −6.814.
the many forms of immobility documented in the NSQIP (quadriparesis, paralysis, hemiparesis, and ventilator dependence).17 Perhaps less anticipated was steroid use as a risk factor. There is some evidence that steroid use is a risk factor for VTE in patients with Crohn’s disease: In one study, 61.5% of patients on steroids developed portal venous thrombosis versus 28.9% of patients not on steroids.11 This association with steroids is also supported by studies showing that patients with Cushing’s disease have an elevated risk of VTE, presumably through a variety of increased procoagulant factors.16 Another underappreciated risk factor identified in this study was sepsis, which has been illustrated outside the neurosurgical context.15 There are a variety of proposed mechanisms for the observed increased thrombogenesis, including overproduction of plasminogen activator inhibitor-1, dysfunctional endothelial cells, and aberrant expression of tissue factor.15 Further study of these pathways in neurosurgical patients may be warranted. A number of limitations to our analysis are worth noting. The NSQIP database documents only VTEs that were symptomatic and required treatment, which probably results in an underrepresentation of the true incidence of VTE since asymptomatic events are not documented. In the same regard, patients with symptomatic VTE who 8
had contraindications to treatment were also probably not included in the analysis. As the NSQIP database collects and tracks data across multiple surgical specialties, the comorbidities and risk adjustment variables that are tracked tend to be general and broad. The presence of cardiac, pulmonary, or endocrinological comorbidities are well documented, as are a number of procedural risk factors, such as whether a case was done in an emergent setting; however, these procedural risk factors are general and not specific to neurosurgery. For example, within spinal surgeries, a single-level laminotomy is included in the same analysis group as a multilevel deformity correction, and it is difficult to differentiate cases by complexity based on variables specific to neurosurgery (for example, levels performed for the same procedure). Another limitation is the lack of information regarding DVT prophylaxis, which is now used more frequently than in the past. The proportion of patients with sequential compression devices, subcutaneous heparin, or low-molecular-weight heparin is unknown, as is the degree to which these interventions lower the rates of VTEs in this patient population. It is our hope that future databases, such as the National Neurosurgery Quality and Outcomes Database (N2QOD), include this and other more granular data to better learn about VTE mitigation strategies and their potential risks in neurosurgical patients. J Neurosurg / August 1, 2014
Pulmonary embolism and deep venous thrombosis in neurosurgery TABLE 5: Significant predictors of DVTs requiring treatment in neurosurgical patients, multivariate model Factor
No. of Patients w/ DVTs Requiring Therapy (%)
surgery type cranial spinal age weight tobacco use no yes ventilator dependence in the 48 hrs before op no yes severe COPD no yes hemiparesis before surgery no yes CNS Tumor no yes paraparesis no yes quadriparesis no yes steroid use no yes chemotherapy no yes sepsis none SIRS sepsis septic shock
OR (95% CI) for Multivariate Model*
259 (2.6) 225 (0.8)
1 [reference] 0.52 (0.39–0.69) 1.02 (1.01–1.03) 1.01 (1.01–1.01)
402 (1.4) 82 (0.9)
1 [reference] 0.70 (0.51–0.96)
431 (1.2) 53 (8.1)
1 [reference] 3.39 (2.19–5.25)
448 (1.2) 36 (2.3)
1 [reference] 1.62 (1.05–2.49)
303 (1.1) 53 (4.5)
1 [reference] 1.81 (1.28–2.55)
245 (1.0) 111 (3.8)
1 [reference] 2.67 (1.97–3.61)
319 (1.2) 37 (3.3)
1 [reference] 2.48 (1.65–3.74)
345 (1.3) 11 (4.7)
1 [reference] 4.44 (2.28–8.68)
417 (1.2) 67 (2.9)
1 [reference] 1.52 (1.08–2.14)
340 (1.2) 16 (5.8)
1 [reference] 2.16 (1.20–3.88)
399 (1.1) 51 (5.7) 21 (7.5) 6 (10.2)
1 [reference] 2.09 (1.39–3.14) 4.26 (2.31–7.85) 3.91 (1.40–3.14)
* Constant for multivariable model was −6.977.
Conclusions Venous thromboembolism is a common complication in neurosurgical patients, having affected 1.7% of patients in the NSQIP database. Deep venous thromboses are roughly twice as common as PEs (1.3% vs 0.6%, respectively), and the rate of VTEs has not changed appreciably over the past several years. Many factors predictive of VTEs were identified in this population: ventilator dependence, immobility (as from quadriparesis, hemiparesis, or J Neurosurg / August 1, 2014
paraparesis), and malignancy. Less anticipated predictors included chronic steroid use and sepsis. Lastly, VTEs appear significantly more likely to occur in patients undergoing cranial procedures (3.4%) than in those undergoing spinal procedures (1.1%). The sources of this discrepancy are unknown but deserve more careful study. It is our hope that a better appreciation of the prevalence and risk factors of VTEs in neurosurgical patients will allow us to better target interventions and have a better understanding of which patients are most at risk. 9
J. D. Rolston et al. TABLE 6: Significant predictors of PE in neurosurgical patients, multivariate results only Factor surgery type cranial spinal age weight altered mental status no yes stroke w/o persistent deficit no yes CNS tumor no yes paraparesis no yes radiotherapy no yes sepsis none SIRS sepsis septic shock
No. of PEs (%)
OR (95% CI)*
137 (1.4) 107 (0.4)
1 [reference] 0.34 (0.22–0.51) 1.03 (1.01–1.04) 1.01 (1.00–1.01)
133 (0.5) 23 (2.2)
1 [reference] 1.93 (1.17–3.18)
154 (0.6) 10 (1.9)
1 [reference] 2.41 (1.19–4.87)
115 (0.5) 49 (1.7)
1 [reference] 1.86 (1.22–2.82)
147 (0.6) 17 (1.5)
1 [reference] 2.78 (1.57–4.90)
157 (0.6) 6 (3.5)
1 [reference] 3.07 (1.20–7.86)
210 (0.6) 25 (2.8) 6 (2.1) 2 (3.4)
1 [reference] 2.69 (1.56–4.63) 2.18 (0.67–7.12) 3.82 (0.86–17.03)
* Constant for multivariable model was −7.849. Disclosure The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper. Dr. Rolston was supported in part by a Socioeconomic Fellowship from the Congress of Neurological Surgeons. The ACS NSQIP and the hospitals participating in the ACS NSQIP are the sources of the data used herein; however, they have not verified and are not responsible for the statistical validity of the data analysis or derived conclusions in this study. Author contributions to the study and manuscript preparation include the following. Conception and design: Parsa, Rolston. Acquisition of data: Rolston, Han. Analysis and interpretation of data: all authors. Drafting the article: all authors. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Parsa. Statistical analysis: Rolston. References 1. Agnelli G, Piovella F, Buoncristiani P, Severi P, Pini M, D’Angelo A, et al: Enoxaparin plus compression stockings compared with compression stockings alone in the prevention of venous thromboembolism after elective neurosurgery. N Engl J Med 339:80–85, 1998 2. Bikdeli B, Sharif-Kashani B: Prophylaxis for venous thromboembolism: a great global divide between expert guidelines and clinical practice? Semin Thromb Hemost 38:144–155, 2012 3. Cerrato D, Ariano C, Fiacchino F: Deep vein thrombosis and
10
low-dose heparin prophylaxis in neurosurgical patients. J Neurosurg 49:378–381, 1978 4. Cheng JS, Arnold PM, Anderson PA, Fischer D, Dettori JR: Anticoagulation risk in spine surgery. Spine (Phila Pa 1976) 35 (9 Suppl):S117–S124, 2010 5. Constantini S, Kanner A, Friedman A, Shoshan Y, Israel Z, Ashkenazi E, et al: Safety of perioperative minidose heparin in patients undergoing brain tumor surgery: a prospective, randomized, double-blind study. J Neurosurg 94:918–921, 2001 6. Hacker RI, Ritter G, Nelson C, Knobel D, Gupta R, Hopkins K, et al: Subcutaneous heparin does not increase postoperative complications in neurosurgical patients: an institutional experience. J Crit Care 27:250–254, 2012 7. Hamilton MG, Yee WH, Hull RD, Ghali WA: Venous thromboembolism prophylaxis in patients undergoing cranial neurosurgery: a systematic review and meta-analysis. Neurosurgery 68:571–581, 2011 8. Hawryluk GW, Furlan JC, Austin JW, Fehlings MG: Survey of neurosurgical management of central nervous system hemorrhage in patients receiving anticoagulation therapy: current practice is highly variable and may be suboptimal. World Neurosurg 76:299–303, 2011 9. Ingraham AM, Richards KE, Hall BL, Ko CY: Quality improvement in surgery: the American College of Surgeons National Surgical Quality Improvement Program approach. Adv Surg 44:251–267, 2010 10. Khuri SF, Henderson WG, Daley J, Jonasson O, Jones RS, Campbell DA Jr, et al: Successful implementation of the Department of Veterans Affairs’ National Surgical Quality Im-
J Neurosurg / August 1, 2014
Pulmonary embolism and deep venous thrombosis in neurosurgery provement Program in the private sector: the Patient Safety in Surgery study. Ann Surg 248:329–336, 2008 11. Leung GG, Sivasankaran MV, Choi JJ, Divino CM: Risk factors of portal vein thrombosis in Crohn’s disease patients. J Gastrointest Surg 16:1199–1203, 2012 12. Nurmohamed MT, van Riel AM, Henkens CM, Koopman MM, Que GT, d’Azemar P, et al: Low molecular weight heparin and compression stockings in the prevention of venous thromboembolism in neurosurgery. Thromb Haemost 75:233–238, 1996 13. Raslan AM, Fields JD, Bhardwaj A: Prophylaxis for venous thrombo-embolism in neurocritical care: a critical appraisal. Neurocrit Care 12:297–309, 2010 14. Rolston JD, Han SJ, Lau CY, Berger MS, Parsa AT: Frequency and predictors of complications in neurological surgery: national trends from 2006 to 2011. Clinical article. J Neurosurg 120:736–745, 2014 15. Semeraro N, Ammollo CT, Semeraro F, Colucci M: Sepsis, thrombosis and organ dysfunction. Thromb Res 129:290–295, 2012
J Neurosurg / August 1, 2014
16. van der Pas R, de Bruin C, Leebeek FW, de Maat MP, Rijken DC, Pereira AM, et al: The hypercoagulable state in Cushing’s disease is associated with increased levels of procoagulant factors and impaired fibrinolysis, but is not reversible after short-term biochemical remission induced by medical therapy. J Clin Endocrinol Metab 97:1303–1310, 2012 17. Wun T, White RH: Venous thromboembolism (VTE) in patients with cancer: epidemiology and risk factors. Cancer Invest 27 (Suppl 1):63–74, 2009
Manuscript submitted July 28, 2013. Accepted June 24, 2014. Please include this information when citing this paper: published online August 1, 2014; DOI: 10.3171/2014.6.JNS131419. Address correspondence to: Andrew T. Parsa, M.D., Ph.D., Department of Neurological Surgery, 400 Parnassus Ave., Box 0350, San Francisco, CA 94143. email: [email protected].
11
What clinical factors predict the incidence of deep venous thrombosis and pulmonary embolism in neurosurgical patients? Clinical article John D. Rolston, M.D., Ph.D.,1 Seunggu J. Han, M.D.,1 Orin Bloch, M.D., 2 and Andrew T. Parsa, M.D., Ph.D. 2 Department of Neurological Surgery, University of California, San Francisco, California; and 2Department of Neurological Surgery, Northwestern University, Chicago, Illinois
1
Object. Venous thromboembolisms (VTEs) occur frequently in surgical patients and can manifest as pulmonary emboli (PEs) or deep venous thromboses (DVTs). While many medical therapies have been shown to prevent VTEs, neurosurgeons are concerned about the use of anticoagulants in the postoperative setting. To better understand the prevalence of and the patient-level risk factors for VTE, the authors analyzed data from the National Surgical Quality Improvement Program (NSQIP). Methods. Retrospective data on 1,777,035 patients for the years from 2006 to 2011 were acquired from the American College of Surgeons NSQIP database. Neurosurgical cases were extracted by querying the data for which the surgical specialty was listed as “neurological surgery.” Univariate statistics were calculated using the chi-square test, with 95% confidence intervals used for the resultant risk ratios. Multivariate models were constructed using binary logistic regression with a maximum number of 20 iterations. Results. Venous thromboembolisms were found in 1.7% of neurosurgical patients, with DVTs roughly twice as common as PEs (1.3% vs 0.6%, respectively). Significant independent predictors included ventilator dependence, immobility (that is, quadriparesis, hemiparesis, or paraparesis), chronic steroid use, and sepsis. The risk of VTE was significantly higher in patients who had undergone cranial procedures (3.4%) than in those who had undergone spinal procedures (1.1%). Conclusions. Venous thromboembolism is a common complication in neurosurgical patients, and the frequency has not changed appreciably over the past several years. Many factors were identified as independently predictive of VTEs in this population: ventilator dependence, immobility, and malignancy. Less anticipated predictors included chronic steroid use and sepsis. Venous thromboembolisms appear significantly more likely to occur in patients undergoing cranial procedures than in those undergoing spinal procedures. A better appreciation of the prevalence of and the risk factors for VTEs in neurosurgical patients will allow targeting of interventions and a better understanding of which patients are most at risk. (http://thejns.org/doi/abs/10.3171/2014.6.JNS131419)
Key Words • pulmonary embolus • deep venous thrombosis • venous thromboembolism • complication • quality improvement • vascular disorders
N
eurosurgical patients are at risk for developing potentially catastrophic deep venous thromboses (DVTs) and pulmonary emboli (PEs).7,13 Unfractionated heparin, low-molecular-weight heparin, compression stockings, and sequential compression devices all re-
Abbreviations used in this paper: ACS = American College of Surgeons; COPD = chronic obstructive pulmonary disease; DVT = deep venous thrombosis; NSQIP = National Surgical Quality Improvement Program; PE = pulmonary embolus; SIRS = systemic inflammatory response syndrome; VTE = venous thromboembolism.
J Neurosurg / August 1, 2014
duce the risk of DVTs and PEs.2,4,7,13 Some of these prophylactic treatments bear specific risks, including intracerebral hemorrhage and bleeding, and concerns about these risks can limit the use of these treatments.2,6–8,13 Here we used data from the American College of Surgeons (ACS) National Surgical Quality Improvement Program (NSQIP)9,10 to identify predictors of developing venous thromboembolism (VTE) and analyzed VTE occurrence over time. Efficient management paradigms will probably depend on identifying those patients most at risk for developing VTE. The long-term goal of the present study is to facilitate a better understanding of which patients will most likely 1
J. D. Rolston et al. benefit from prophylactic treatments that mean added costs to health care while potentially incurring risk.
Methods
Data on 1,777,035 patients for the years 2006 to 2011 were acquired from the ACS NSQIP database.9,10 This database contains a nationwide data set representing cases from a wide range of academic and private hospitals across 47 US states.14 Cases are analyzed for a predetermined collection of complications, including VTEs. Reviewers are frequently audited, and stringent criteria for each complication are used. In particular, DVTs must be diagnosed within 30 days of an operation by using duplex ultrasound, venography, or CT scanning, and patients must be treated for these clots with an inferior vena cava filter, inferior vena cava ligation, or anticoagulation therapy. Pulmonary embolisms must occur within 30 days of an operation and must be confirmed with a highprobability ventilation-perfusion (VQ) scan, CT scan, or pulmonary arteriogram. Database files were acquired in delimited text format and parsed using both SPSS version 20 (IBM Corp.) and MATLAB R2012a (MathWorks Inc.). Neurosurgical cases were extracted by querying the data for which the surgical specialty was listed as “neurological surgery” and further classified as “spine” or “cranial” depending on the current procedural terminology (CPT) code (Table 1). Descriptive statistics are represented as the means ± standard deviation. Univariate statistics were calculated using the chi-square test, and 95% confidence intervals are presented for the resultant risk ratios. Predictors included in the models are listed in Table 2. Multivariate models were constructed using binary logistic regression with a maximum of 20 iterations. Variables that were significant (p < 0.05) on univariate analysis were included in the multivariate model. During model building, insignificant predictors (p > 0.05) were removed in a stepwise fashion using the Wald statistic. Significant predictors from this model were identified, and their odds ratios were presented along with the 95% confidence intervals.
Results Demographics and Prevalence
Using the NSQIP database, we identified 38,058 neurosurgical cases for the years 2006 to 2011. Patient demographics are shown in Table 3. Six hundred forty cases of VTE were identified, corresponding to 1.7% of the neurosurgical cases. Of these cases, 244 (0.6%) involved PEs and 484 (1.3%) involved DVTs requiring treatment. Asymptomatic and untreated DVTs were not included in this analysis. The prevalence of VTEs was consistent from year to year, ranging from a low of 1.3% in 2007 to a high of 1.8% in 2008. There was no significant trend in the percentage of VTEs over time (F = 0.759; p = 0.45). This consistency in prevalence was also true for DVTs (min 1.0% in 2007 and max 1.5% in 2008; F = 0.305; p = 0.62) and PEs (min 0.4% in 2007 and max 0.7% in 2009; F = 0.304; p = 0.62).
2
TABLE 1: Current procedural terminology (CPT) codes included in study Spine Codes
Cranial Codes
0090T 0092T 0093T 0095T 0096T 0163T 0195T 0196T 0202T 10121 10140 10180 11000 11042-4 20005 20100 20696 20902 20922 20930 20931 20936 20938 20999 21026 21501 21510 21610 21615 22010 22015 22100-14 22206-26 22318-27 22520-95 22600-32 22800-99 27080 27202 27218 27280 42725 49215 63001-308 63685 63700-41 64573-85
21137 21139 21180 21299 35001-5 35201 35301 35800 37605 37799 49422 60600 61150-888 62000-362 69511
J Neurosurg / August 1, 2014
Pulmonary embolism and deep venous thrombosis in neurosurgery TABLE 2: Predictors included in models* Patient Characteristic sex race type of surgery age year of operation height weight diabetes tobacco use ethanol use dyspnea ventilator dependence in 48 hrs prior to surgery severe COPD preop pneumonia ascites esophageal varices congestive heart failure recent MI prior PCI prior cardiac surgery angina hypertension requiring medication peripheral vascular disease rest pain renal failure dialysis altered mental status coma hemiparesis TIA stroke w/ persistent deficit stroke w/o persistent deficit CNS tumor paraparesis quadriparesis disseminated cancer wound infection chronic steroid use recent weight loss bleeding disorder transfusion chemotherapy radiotherapy sepsis
Value M, F white, African American, Asian or Pacific Islander, American Indian or Alaskan Native cranial, spinal 16–90 yrs 2006–2011
none, insulin independent, insulin dependent no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes no, yes none, SIRS, sepsis, septic shock
* MI = myocardial infarction; PCI = percutaneous coronary intervention; TIA = transient ischemic attack.
J Neurosurg / August 1, 2014
TABLE 3: Basic demographics for 38,058 neurosurgical patients in the NSQIP database* Characteristic
No. (%)
mean age in yrs ± SD mean weight in lbs ± SD mean height in inches ± SD sex M F unknown race white African American Asian or Pacific Islander American Indian or Alaskan Native unknown surgery type cranial spinal unknown inpatient vs outpatient inpatient outpatient unknown diabetes no yes insulin dependent not insulin dependent unknown tobacco use no yes unknown ethanol use no yes unknown ventilator dependence in 48 hrs before op no yes unknown severe COPD no yes unknown preop pneumonia no yes unknown
56.0 ± 15.0 186.7 ± 46.9 66.9 ± 4.2 19,349 (50.8) 18,634 (49.0) 75 (0.2) 30,104 (79.1) 2922 (7.7) 722 (1.9) 221 (0.6) 4089 (10.7) 10,041 (26.4) 28,017 (73.6) 0 32,628 (85.7) 5430 (14.3) 0 32,630 (85.7) 5428 (14.3) 1906 (5.0) 3522 (9.3) 0 28,434 (74.7) 9623 (25.3) 1 (0.0) 26,530 (69.7) 1064 (2.8) 10,464 (27.5) 37,401 (98.3) 657 (1.7) 0 36,507 (95.9) 1551 (4.1) 0 27,505 (72.3) 69 (0.2) 10,484 (27.5) (continued)
3
J. D. Rolston et al. TABLE 3: Basic demographics for 38,058 neurosurgical patients in the NSQIP database* (continued) Characteristic ascites no yes unknown esophageal varices no yes unknown CHF no yes unknown prior MI no yes unknown prior PCI no yes unknown prior cardiac surgery no yes unknown angina no yes unknown hypertension requiring medication no yes unknown peripheral vascular disease no yes unknown rest pain no yes unknown renal failure no yes unknown dialysis no yes unknown
No. (%) 38,037 (99.9) 21 (0.1) 0 27,566 (72.4) 8 (0.0) 10,484 (27.5) 37,931 (99.7) 127 (0.3) 0 27,497 (72.3) 77 (0.2) 10,484 (27.5) 26,173 (68.8) 1401 (3.7) 10,484 (27.5) 26,450 (69.5) 1123 (3.0) 10,485 (27.6) 27,430 (72.1) 143 (0.4) 10,485 (27.6) 20,205 (53.1) 17,853 (46.9) 0 27,311 (71.8) 263 (0.7) 10,484 (27.5) 27,521 (72.3) 52 (0.1) 10,485 (27.6) 38,003 (99.9) 54 (0.1) 1 (0.0) 37,899 (99.6) 158 (0.4) 1 (0.0) (continued)
4
TABLE 3: Basic demographics for 38,058 neurosurgical patients in the NSQIP database* (continued) Characteristic
No. (%)
altered mental status no yes unknown coma no yes unknown hemiparesis before surgery no yes unknown prior TIA no yes unknown history of stroke w/ persistent deficit no yes unknown history of stroke w/o persistent deficit no yes unknown CNS tumor no yes unknown paraparesis no yes unknown quadriparesis no yes unknown disseminated cancer no yes unknown wound infection no yes unknown steroid use no yes unknown
26,524 (69.7) 1050 (2.8) 10,484 (27.5) 27,350 (71.9) 224 (0.6) 10,484 (27.5) 26,396 (69.4) 1178 (3.1) 10,484 (27.5) 26,918 (70.7) 656 (1.7) 10,484 (27.5) 26,441 (69.5) 1133 (3.0) 10,484 (27.5) 27,048 (71.1) 526 (1.4) 10,484 (27.5) 24,669 (64.8) 2905 (7.6) 10,484 (27.5) 26,463 (69.5) 1111 (2.9) 10,484 (27.5) 27,339 (71.8) 235 (0.6) 10,484 (27.5) 36,527 (96.0) 1531 (4.0) 0 37,427 (98.3) 631 (1.7) 0 35,740 (93.9) 2318 (6.1) 0 (continued)
J Neurosurg / August 1, 2014
Pulmonary embolism and deep venous thrombosis in neurosurgery TABLE 3: Basic demographics for 38,058 neurosurgical patients in the NSQIP database* (continued) Characteristic recent weight loss no yes unknown bleeding disorder no yes unknown transfusion no yes unknown chemotherapy no yes unknown radiotherapy no yes unknown sepsis none SIRS sepsis septic shock unknown emergency case no yes
No. (%) 37,633 (98.9) 425 (1.1) 0 36,966 (97.1) 1091 (2.9) 1 (0.0) 37,913 (99.6) 145 (0.4) 0 27,300 (71.7) 274 (0.7) 10,484 (27.5) 27,218 (71.5) 173 (0.5) 10,667 (28.0 36,581 (96.1) 896 (2.4) 281 (0.7) 59 (0.2) 241 (0.6) 35,806 (94.1) 2252 (5.9)
* CHF = congestive heart failure.
Predictors of VTE
Multivariate statistical analysis was used to identify patient characteristics predictive of VTE, PE, and DVT formation. For VTEs (that is, DVTs requiring treatment and PEs), the rate of cranial cases was far higher than the rate of spinal cases (3.4% vs 1.1%; Table 4). This difference was significant, with an OR of 0.44 (95% CI 0.34– 0.57), showing the lower likelihood of a VTE developing in spinal patients as compared with cranial patients. This difference persisted when analyzing solely DVTs or PEs. For DVTs alone, there were events in 259 cranial cases (2.6%) versus 225 spinal cases (0.8%; OR 0.52, 95% CI 0.39–0.69; Table 5). For PEs alone, there were events in 137 cranial cases (1.4%) as compared with 107 spinal cases (0.4%; OR 0.34, 95% CI 0.22–0.51; Table 6). Other significant predictors of VTE included the presence of a CNS tumor (OR 2.24, 95% CI 1.71–2.94), any significant motor weakness (whether hemiparesis [OR 1.80, 95% CI 1.32–2.45], quadriparesis [OR 3.78,
J Neurosurg / August 1, 2014
95% CI 2.00–7.17] or paraparesis [OR 2.56, 95% CI 1.78– 3.69]), chronic steroid use (OR 1.63, 95% CI 1.20–2.21), and ventilator dependence prior to surgery (OR 2.41, 95% CI 1.60–3.62; Table 4). Other notable predictors were the presence of systemic inflammatory response syndrome (SIRS) criteria (OR 2.32, 95% CI 1.61–3.32), sepsis (OR 3.97, 95% CI 2.25–7.02), or septic shock (OR 3.58, 95% CI 1.32–9.81). These individual predictors were also identified when constructing a multivariate model exclusively with DVTs (Table 5). However, three additional significant protective predictors were found: severe chronic obstructive pulmonary disease (COPD; OR 1.62, 95% CI 1.05–2.49), chemotherapy use (OR 2.16, 95% CI 1.20–3.88), and tobacco use (OR 0.70, 95% CI 0.51–0.96). Because there were fewer PEs than DVTs (244 vs 484) and consequently less power in the statistical analysis, fewer significant predictors were identified in the analysis of PEs when examined in isolation (Table 6). Nevertheless, the type of procedure (cranial vs spinal) remained significant, as did the presence of paraparesis (OR 2.78, 95% CI 1.57–4.90) or SIRS (OR 2.69, 95% CI 1.56–4.63). One additional predictor, not found in the analysis of either VTEs or DVTs, was an altered mental status (OR 1.93, 95% CI 1.17–3.18). Again, these predictors were identified even when confounding variables, such as the presence of disseminated cancer or CNS tumors, were controlled for.
Discussion
Using the NSQIP database, which contains data from more than 1 million surgical patients, we analyzed patient-level predictors of VTE and, individually, of DVT and PE. Patients who underwent spine surgery were significantly less likely to develop DVTs or PEs than those who underwent cranial surgery, with an OR of 0.44 (95% CI 0.34–0.57). This was true even when accounting for multiple confounders using multivariate logistic regression (see Table 2 for a list of variables). The cause of this difference is unknown. Perhaps the NSQIP data set is lacking important predictors of VTE that would otherwise account for the discrepancy, although a great deal of variables are included, such as the presence of a CNS tumor, paralysis, stroke, coma, ventilator dependence, and so forth (Table 3). Alternatively, physician practice could explain some of the difference. Perhaps cranial surgeons are less likely than spinal surgeons to prescribe VTE prophylaxis. If so, this tendency could be attributable to a heightened concern among cranial surgeons for postoperative bleeding and the devastating consequences thereof, despite the fact that multiple studies refute this idea.1,3,5,7,12 It should be noted that the NSQIP database contains no information about whether thromboprophylaxis was used. A last possibility relates to follow-up time. Perhaps cranial patients are hospitalized longer and thus have more time for VTEs to present themselves in the hospital setting and to show up as complications within the medical record. Other predictors of VTE identified in this study have been previously mentioned and thus offer validity to the methodology used herein. These predictors include the presence of cancer and venostasis, such as that found in 5
J. D. Rolston et al. TABLE 4: Significant predictors of VTE in neurosurgical patients, both univariate and multivariate Factor age weight surgery type cranial spinal hemiparesis before surgery no yes history of stroke w/o persistent deficit no yes chemotherapy no yes ventilator dependence in 48 hrs before op no yes CNS tumor no yes paraparesis no yes quadriparesis no yes steroid use no yes sepsis none SIRS sepsis septic shock severe COPD no yes tobacco use no yes patient sex M F race white African American Asian or Pacific Islander American Indian or Alaskan Native
No. of VTEs (%)
Univariate OR (95% CI)
Multivariate OR (95% CI)† 1.03 (1.02–1.03)* 1.01 (1.01–1.01)*
345 (3.4) 295 (1.1)
1 [reference] 0.30 (0.26–0.35)*
1 [reference] 0.44 (0.34–0.57)*
387 (1.5) 70 (5.9)
1 [reference] 4.25 (3.27–5.52)*
1 [reference] 1.80 (1.32–2.45)*
437 (1.6) 20 (3.8)
1 [reference] 2.41 (1.52–3.80)*
1 [reference] 1.85 (1.12–3.05)*
437 (1.6) 20 (7.3)
1 [reference] 4.84 (3.04–7.71)*
1 [reference] 1.94 (1.12–3.36)*
581 (1.6) 59 (9.0)
1 [reference] 6.25 (4.73–8.27)*
1 [reference] 2.41 (1.60–3.62)*
322 (1.3) 135 (4.6)
1 [reference] 3.69 (3.00–4.52)*
1 [reference] 2.24 (1.71–2.94)*
412 (1.6) 45 (4.1)
1 [reference] 2.67 (1.95–3.65)*
1 [reference] 2.56 (1.78–3.69)*
445 (1.6) 12 (5.1)
1 [reference] 3.25 (1.81–5.86)*
1 [reference] 3.78 (2.00–7.17)*
550 (1.5) 90 (3.9)
1 [reference] 2.58 (2.06–3.24)*
1 [reference] 1.63 (1.20–2.21)*
538 (1.5) 64 (7.1) 24 (8.5) 6 (10.2)
1 [reference] 5.15 (3.94–6.74)* 6.26 (4.08–9.59)* 7.58 (3.25–17.72)*
1 [reference] 2.32 (1.61–3.32)* 3.97 (2.25–7.02)* 3.58 (1.32–9.81)*
592 (1.6) 48 (3.1)
1 [reference] 1.94 (1.44–2.61)*
532 (1.9) 108 (1.1)
1 [reference] 0.60 (0.48–0.73)*
353 (1.8) 286 (1.5)
1 [reference] 0.84 (0.72–0.98)*
487 (1.6) 81 (2.8) 8 (1.1) 3 (1.4)
1 [reference] 1.74 (1.37–2.20)* 0.68 (0.34–1.38) 0.84 (0.27–2.62) (continued)
6
J Neurosurg / August 1, 2014
Pulmonary embolism and deep venous thrombosis in neurosurgery TABLE 4: Significant predictors of VTE in neurosurgical patients, both univariate and multivariate (continued) Factor ethanol use no yes hypertension requiring medication no yes preop pneumonia no yes dialysis no yes renal failure no yes altered mental status no yes coma no yes history of TIA no yes history of stroke w/ persistent deficit no yes disseminated cancer no yes wound infection no yes recent weight loss no yes bleeding disorder no yes transfusion no yes radiotherapy no yes CHF no yes
No. of VTEs (%)
Univariate OR (95% CI)
443 (1.7) 14 (1.3)
1 [reference] 0.79 (0.46–1.34)
279 (1.4) 361 (2.0)
1 [reference] 1.47 (1.26–1.73)*
453 (1.7) 4 (5.8)
1 [reference] 3.67 (1.33–10.13)*
638 (1.7) 2 (1.3)
1 [reference] 0.75 (0.19–3.03)
637 (1.7) 3 (5.6)
1 [reference] 3.45 (1.07–11.08)*
383 (1.4) 74 (7.0
1 [reference] 5.17 (4.00–6.69)*
438 (1.6) 19 (8.5)
1 [reference] 5.69 (3.53–9.20)*
438 (1.6) 19 (2.9)
1 [reference] 1.80 (1.13–2.87)*
394 (1.5) 63 (5.6)
1 [reference] 3.89 (2.96–5.11)*
575 (1.6) 65 (4.2)
1 [reference] 2.77 (2.13–3.60)*
616 (1.6) 24 (3.8)
1 [reference] 2.36 (1.56–3.58)*
629 (1.7) 11 (2.6)
1 [reference] 1.56 (0.85–2.86)
592 (1.6) 48 (4.4)
1 [reference] 2.83 (2.09–3.82)*
632 (1.7) 8 (5.5)
1 [reference] 3.44 (1.68–7.06)*
441 (1.6) 12 (6.9)
1 [reference] 4.53 (2.50–8.20)*
635 (1.7) 5 (3.9)
1 [reference] 2.41 (0.98–5.91)
Multivariate OR (95% CI)†
(continued)
J Neurosurg / August 1, 2014
7
J. D. Rolston et al. TABLE 4: Significant predictors of VTE in neurosurgical patients, both univariate and multivariate (continued) Factor
No. of VTEs (%)
Univariate OR (95% CI)
455 (1.7) 2 (2.6)
1 [reference] 1.58 (0.39–6.47)
436 (1.7) 21 (1.5)
1 [reference] 0.90 (0.58–1.40)
425 (1.6) 31 (2.8)
1 [reference] 1.74 (1.20–2.56)*
452 (1.6) 4 (2.8)
1 [reference] 1.72 (0.63–4.67)
451 (1.7) 6 (2.3)
1 [reference] 1.39 (0.62–3.14)
455 (1.7) 1 (1.9)
1 [reference] 1.17 (0.16–8.46)
456 (1.7) 1 (12.5)
1 [reference] 8.49 (1.04–69.17)*
prior MI no yes prior PCI no yes prior cardiac surgery no yes angina no yes peripheral vascular disease no yes rest pain no yes esophageal varices no yes ascites no yes
639 (1.7) 1 (4.8)
Multivariate OR (95% CI)†
1 [reference] 2.93 (0.39–21.84)
* Indicates a statistically significant odds ratio. † Constant for multivariable model was −6.814.
the many forms of immobility documented in the NSQIP (quadriparesis, paralysis, hemiparesis, and ventilator dependence).17 Perhaps less anticipated was steroid use as a risk factor. There is some evidence that steroid use is a risk factor for VTE in patients with Crohn’s disease: In one study, 61.5% of patients on steroids developed portal venous thrombosis versus 28.9% of patients not on steroids.11 This association with steroids is also supported by studies showing that patients with Cushing’s disease have an elevated risk of VTE, presumably through a variety of increased procoagulant factors.16 Another underappreciated risk factor identified in this study was sepsis, which has been illustrated outside the neurosurgical context.15 There are a variety of proposed mechanisms for the observed increased thrombogenesis, including overproduction of plasminogen activator inhibitor-1, dysfunctional endothelial cells, and aberrant expression of tissue factor.15 Further study of these pathways in neurosurgical patients may be warranted. A number of limitations to our analysis are worth noting. The NSQIP database documents only VTEs that were symptomatic and required treatment, which probably results in an underrepresentation of the true incidence of VTE since asymptomatic events are not documented. In the same regard, patients with symptomatic VTE who 8
had contraindications to treatment were also probably not included in the analysis. As the NSQIP database collects and tracks data across multiple surgical specialties, the comorbidities and risk adjustment variables that are tracked tend to be general and broad. The presence of cardiac, pulmonary, or endocrinological comorbidities are well documented, as are a number of procedural risk factors, such as whether a case was done in an emergent setting; however, these procedural risk factors are general and not specific to neurosurgery. For example, within spinal surgeries, a single-level laminotomy is included in the same analysis group as a multilevel deformity correction, and it is difficult to differentiate cases by complexity based on variables specific to neurosurgery (for example, levels performed for the same procedure). Another limitation is the lack of information regarding DVT prophylaxis, which is now used more frequently than in the past. The proportion of patients with sequential compression devices, subcutaneous heparin, or low-molecular-weight heparin is unknown, as is the degree to which these interventions lower the rates of VTEs in this patient population. It is our hope that future databases, such as the National Neurosurgery Quality and Outcomes Database (N2QOD), include this and other more granular data to better learn about VTE mitigation strategies and their potential risks in neurosurgical patients. J Neurosurg / August 1, 2014
Pulmonary embolism and deep venous thrombosis in neurosurgery TABLE 5: Significant predictors of DVTs requiring treatment in neurosurgical patients, multivariate model Factor
No. of Patients w/ DVTs Requiring Therapy (%)
surgery type cranial spinal age weight tobacco use no yes ventilator dependence in the 48 hrs before op no yes severe COPD no yes hemiparesis before surgery no yes CNS Tumor no yes paraparesis no yes quadriparesis no yes steroid use no yes chemotherapy no yes sepsis none SIRS sepsis septic shock
OR (95% CI) for Multivariate Model*
259 (2.6) 225 (0.8)
1 [reference] 0.52 (0.39–0.69) 1.02 (1.01–1.03) 1.01 (1.01–1.01)
402 (1.4) 82 (0.9)
1 [reference] 0.70 (0.51–0.96)
431 (1.2) 53 (8.1)
1 [reference] 3.39 (2.19–5.25)
448 (1.2) 36 (2.3)
1 [reference] 1.62 (1.05–2.49)
303 (1.1) 53 (4.5)
1 [reference] 1.81 (1.28–2.55)
245 (1.0) 111 (3.8)
1 [reference] 2.67 (1.97–3.61)
319 (1.2) 37 (3.3)
1 [reference] 2.48 (1.65–3.74)
345 (1.3) 11 (4.7)
1 [reference] 4.44 (2.28–8.68)
417 (1.2) 67 (2.9)
1 [reference] 1.52 (1.08–2.14)
340 (1.2) 16 (5.8)
1 [reference] 2.16 (1.20–3.88)
399 (1.1) 51 (5.7) 21 (7.5) 6 (10.2)
1 [reference] 2.09 (1.39–3.14) 4.26 (2.31–7.85) 3.91 (1.40–3.14)
* Constant for multivariable model was −6.977.
Conclusions Venous thromboembolism is a common complication in neurosurgical patients, having affected 1.7% of patients in the NSQIP database. Deep venous thromboses are roughly twice as common as PEs (1.3% vs 0.6%, respectively), and the rate of VTEs has not changed appreciably over the past several years. Many factors predictive of VTEs were identified in this population: ventilator dependence, immobility (as from quadriparesis, hemiparesis, or J Neurosurg / August 1, 2014
paraparesis), and malignancy. Less anticipated predictors included chronic steroid use and sepsis. Lastly, VTEs appear significantly more likely to occur in patients undergoing cranial procedures (3.4%) than in those undergoing spinal procedures (1.1%). The sources of this discrepancy are unknown but deserve more careful study. It is our hope that a better appreciation of the prevalence and risk factors of VTEs in neurosurgical patients will allow us to better target interventions and have a better understanding of which patients are most at risk. 9
J. D. Rolston et al. TABLE 6: Significant predictors of PE in neurosurgical patients, multivariate results only Factor surgery type cranial spinal age weight altered mental status no yes stroke w/o persistent deficit no yes CNS tumor no yes paraparesis no yes radiotherapy no yes sepsis none SIRS sepsis septic shock
No. of PEs (%)
OR (95% CI)*
137 (1.4) 107 (0.4)
1 [reference] 0.34 (0.22–0.51) 1.03 (1.01–1.04) 1.01 (1.00–1.01)
133 (0.5) 23 (2.2)
1 [reference] 1.93 (1.17–3.18)
154 (0.6) 10 (1.9)
1 [reference] 2.41 (1.19–4.87)
115 (0.5) 49 (1.7)
1 [reference] 1.86 (1.22–2.82)
147 (0.6) 17 (1.5)
1 [reference] 2.78 (1.57–4.90)
157 (0.6) 6 (3.5)
1 [reference] 3.07 (1.20–7.86)
210 (0.6) 25 (2.8) 6 (2.1) 2 (3.4)
1 [reference] 2.69 (1.56–4.63) 2.18 (0.67–7.12) 3.82 (0.86–17.03)
* Constant for multivariable model was −7.849. Disclosure The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper. Dr. Rolston was supported in part by a Socioeconomic Fellowship from the Congress of Neurological Surgeons. The ACS NSQIP and the hospitals participating in the ACS NSQIP are the sources of the data used herein; however, they have not verified and are not responsible for the statistical validity of the data analysis or derived conclusions in this study. Author contributions to the study and manuscript preparation include the following. Conception and design: Parsa, Rolston. Acquisition of data: Rolston, Han. Analysis and interpretation of data: all authors. Drafting the article: all authors. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Parsa. Statistical analysis: Rolston. References 1. Agnelli G, Piovella F, Buoncristiani P, Severi P, Pini M, D’Angelo A, et al: Enoxaparin plus compression stockings compared with compression stockings alone in the prevention of venous thromboembolism after elective neurosurgery. N Engl J Med 339:80–85, 1998 2. Bikdeli B, Sharif-Kashani B: Prophylaxis for venous thromboembolism: a great global divide between expert guidelines and clinical practice? Semin Thromb Hemost 38:144–155, 2012 3. Cerrato D, Ariano C, Fiacchino F: Deep vein thrombosis and
10
low-dose heparin prophylaxis in neurosurgical patients. J Neurosurg 49:378–381, 1978 4. Cheng JS, Arnold PM, Anderson PA, Fischer D, Dettori JR: Anticoagulation risk in spine surgery. Spine (Phila Pa 1976) 35 (9 Suppl):S117–S124, 2010 5. Constantini S, Kanner A, Friedman A, Shoshan Y, Israel Z, Ashkenazi E, et al: Safety of perioperative minidose heparin in patients undergoing brain tumor surgery: a prospective, randomized, double-blind study. J Neurosurg 94:918–921, 2001 6. Hacker RI, Ritter G, Nelson C, Knobel D, Gupta R, Hopkins K, et al: Subcutaneous heparin does not increase postoperative complications in neurosurgical patients: an institutional experience. J Crit Care 27:250–254, 2012 7. Hamilton MG, Yee WH, Hull RD, Ghali WA: Venous thromboembolism prophylaxis in patients undergoing cranial neurosurgery: a systematic review and meta-analysis. Neurosurgery 68:571–581, 2011 8. Hawryluk GW, Furlan JC, Austin JW, Fehlings MG: Survey of neurosurgical management of central nervous system hemorrhage in patients receiving anticoagulation therapy: current practice is highly variable and may be suboptimal. World Neurosurg 76:299–303, 2011 9. Ingraham AM, Richards KE, Hall BL, Ko CY: Quality improvement in surgery: the American College of Surgeons National Surgical Quality Improvement Program approach. Adv Surg 44:251–267, 2010 10. Khuri SF, Henderson WG, Daley J, Jonasson O, Jones RS, Campbell DA Jr, et al: Successful implementation of the Department of Veterans Affairs’ National Surgical Quality Im-
J Neurosurg / August 1, 2014
Pulmonary embolism and deep venous thrombosis in neurosurgery provement Program in the private sector: the Patient Safety in Surgery study. Ann Surg 248:329–336, 2008 11. Leung GG, Sivasankaran MV, Choi JJ, Divino CM: Risk factors of portal vein thrombosis in Crohn’s disease patients. J Gastrointest Surg 16:1199–1203, 2012 12. Nurmohamed MT, van Riel AM, Henkens CM, Koopman MM, Que GT, d’Azemar P, et al: Low molecular weight heparin and compression stockings in the prevention of venous thromboembolism in neurosurgery. Thromb Haemost 75:233–238, 1996 13. Raslan AM, Fields JD, Bhardwaj A: Prophylaxis for venous thrombo-embolism in neurocritical care: a critical appraisal. Neurocrit Care 12:297–309, 2010 14. Rolston JD, Han SJ, Lau CY, Berger MS, Parsa AT: Frequency and predictors of complications in neurological surgery: national trends from 2006 to 2011. Clinical article. J Neurosurg 120:736–745, 2014 15. Semeraro N, Ammollo CT, Semeraro F, Colucci M: Sepsis, thrombosis and organ dysfunction. Thromb Res 129:290–295, 2012
J Neurosurg / August 1, 2014
16. van der Pas R, de Bruin C, Leebeek FW, de Maat MP, Rijken DC, Pereira AM, et al: The hypercoagulable state in Cushing’s disease is associated with increased levels of procoagulant factors and impaired fibrinolysis, but is not reversible after short-term biochemical remission induced by medical therapy. J Clin Endocrinol Metab 97:1303–1310, 2012 17. Wun T, White RH: Venous thromboembolism (VTE) in patients with cancer: epidemiology and risk factors. Cancer Invest 27 (Suppl 1):63–74, 2009
Manuscript submitted July 28, 2013. Accepted June 24, 2014. Please include this information when citing this paper: published online August 1, 2014; DOI: 10.3171/2014.6.JNS131419. Address correspondence to: Andrew T. Parsa, M.D., Ph.D., Department of Neurological Surgery, 400 Parnassus Ave., Box 0350, San Francisco, CA 94143. email: [email protected].
11
