Abstract
Background
This study examined outcomes of endovascular repair of infrarenal abdominal aortic aneurysms (EVAR) using general, spinal, epidural, and local/monitored anesthesia care (MAC) in a multicenter North American hospital database reflecting contemporary anesthesia and surgical practices.
Methods
Elective EVAR cases performed between 2005 and 2008 were identified from the American College of Surgeons National Surgical Quality Improvement Program database using Current Procedural Terminology codes. Excluded were emergency cases and patients with concomitant procedures requiring general anesthesia. Patient-level comorbidities, characteristics, and intraoperative and postoperative details were examined. Complications were analyzed individually and in aggregate categories, including wound, pulmonary, renal, venous thromboembolic, cardiovascular, operative, and septic. Length of stay (LOS) and 30-day mortality were examined. Characteristics and outcomes were described using mean ± standard deviation or count (%), and comparisons were evaluated for statistical significance using χ2, Fisher exact test, and univariate linear regression. LOS was analyzed with linear regression techniques using a log transformation. Associations between anesthesia type and outcomes were examined using univariable and multivariable regression techniques.
Results
We identified 6009 elective EVAR procedures for analysis. General anesthesia was used in 4868 cases, spinal anesthesia in 419, epidural anesthesia in 331, and local/MAC in 391. Defined morbidity occurred in 11% of patients. Median LOS was 2 (interquartile range, 1–3) days, and mean LOS was 2.8 ± 4.3 days. The 30-day mortality rate was 1.1%. Significant multivariate associations were observed between anesthesia type, pulmonary morbidity, and log-LOS. General anesthesia was associated with an increase in pulmonary morbidity vs spinal (odds ratio [OR], 4.0; 95% confidence interval [CI], 1.3–12.5; P = .020) and local/MAC anesthesia (OR, 2.6; 95% CI, 1.0–6.4; P = .041). Use of general anesthesia was associated with a 10% increase in LOS for general vs spinal anesthesia (95% CI, 4.8%–15.5%; P = .001) and a 20% increase for general vs local/MAC anesthesia (95% CI, 14.1%–26.2%; P < .001). Trends toward increased pulmonary morbidity and LOS were not observed for general vs epidural anesthesia. No significant association between anesthesia type and mortality was observed.
Conclusions
In contemporary North American anesthetic and surgical practice, general anesthesia for EVAR was associated with increased postoperative LOS and pulmonary morbidity compared with spinal and local/MAC anesthesia. These data suggest that increasing the use of less-invasive anesthetic techniques may limit postoperative complications and decrease the overall costs of EVAR.
Endovascular repair of abdominal aortic aneurysms (EVAR) was introduced in 1990 with the goal of offering a lower-risk alternative to traditional open surgical repair.1 Over time, EVAR has been proven to reduce certain classes of morbidity and hospital length of stay (LOS), with conflicting results regarding reductions in early-term and long-term mortality rates.2,3 Significant rates of cardiac, renal, wound-related, and pulmonary morbidity still occur with EVAR due to the relatively high-risk population inherent with aneurysmal disease of the aorta.4 Surgical teams interested in minimizing these complications have sought to capitalize on the less-invasive nature of EVAR and limit perioperative morbidity in several ways, including the use of alternative anesthesia strategies.
Various anesthetic techniques can be applied to successfully accomplish EVAR, including general anesthesia, regional anesthesia (including epidural and spinal anesthesia), and local anesthesia, with or without monitored anesthesia care (MAC).5,6 Single-center and multicenter reports have examined the results with these various anesthetic techniques for EVAR, with varying results.7–11 This study examined the results of various anesthetic techniques for EVAR in contemporary North American anesthetic and surgical practice by using a large, multicenter data source, the American College of Surgeons National Surgical Quality Improvement (ACS NSQIP) database.
METHODS
Data source
The ACS NSQIP is a validated, prospective database derived from a systematic sampling of cases at 211 participating hospitals throughout North America. Available data include patient demographics, medical risk factors, and detailed information regarding procedural specifics and postoperative morbidity and mortality. All data are collected at participating sites by trained research nurses. Definitions for the variables collected in the NSQIP database have been described in previous reports.12–14
Study sample
Elective EVAR procedures performed between January 2005 and December 2008 were identified by querying the ACS NSQIP for cases with the use of Current Procedural Terminology (CPT) codes (American Medical Association, Chicago, Ill) for the deployment of the main body of an endovascular aortic stent graft (CPT codes 34800, 34802, 34803, 34804, 34805). The study excluded cases coded as emergencies with International Classification of Disease, 9th Edition code 441.3, which designates a ruptured abdominal aortic aneurysm, cases including codes for intraoperative open surgical conversion (CPT codes 34830–34832), cases involving an iliac artery exposure or conduit creation (CPT codes 34820 and 34833), and cases involving concomitant operative procedures requiring general anesthesia. This was done to minimize biases introduced from the analysis of cases in which anesthetic choices were limited or crossovers occurred between anesthetic techniques, or both. This sampling strategy resulted in the identification of 6009 elective EVAR cases for analysis.
Demographics and medical risk factors
All demographic and medical risk factor data were extracted directly from the ACS NSQIP database. Race was considered as white or nonwhite (including Hispanic, Asian, Native American, and black).
Age was considered as a continuous variable for the purpose of this analysis. The ACS NSQIP public-use file database contains a numeric age in years for all records but codes all individuals aged >90 years as 90 years to prevent the potential identification of individual patients during analysis. In the study sample for this analysis, 110 of 6009 patients (1.8%) were coded as having an age of 90 years.
American Society of Anesthesiology (ASA) classes were considered as ASA category 1, 2, and 3 compared with ASA classes 4 and 5 in the multivariable models owing to the relative paucity of patients in ASA classes 1, 2, and 5.
The estimated glomerular filtration rate (eGFR) was used to assess renal function and was calculated using the abbreviated Modification of Diet in Renal Disease formula.15 Body mass index (BMI) was calculated using weight and height data (kg/m2). Operative time was defined in the NSQIP data as the total operation time in minutes.
Anesthesia type
Data regarding the type of anesthetic was extracted from the ACS NSQIP for all identified elective EVAR cases. Anesthesia type was designated in the ACS NSQIP database as general, epidural, spinal, local, MAC, and other. All cases designated as “local” or “MAC” were combined (local/MAC) because some form of local analgesia is required for the EVAR procedure even in the presence of centrally acting sedative and dissociative agents. This analysis excluded 15 cases coded as “other,” one coded as “none,” and 36 coded as “regional.”
End points
Three major outcomes were analyzed for the purposes of this investigation: morbidity, mortality, and length of stay. Postoperative complications (morbidity) were analyzed individually, and in aggregate categories, including:
wound: superficial or deep surgical site infections;
pulmonary: pneumonia, reintubation, or failure to wean from ventilator ≤48 hours from the end time of the surgical procedure;
renal: postoperative renal function decline or need for dialysis;
venous thromboembolic: deep vein thrombosis or pulmonary embolism;
cardiovascular: myocardial infarction, cardiac arrest, or stroke;
operative: return to operating room, postoperative bleeding, or graft failure; and
septic: sepsis and septic shock.
Postoperative mortality was defined as death ≤30 days or during the same acute-care hospital stay, regardless of time. LOS was defined as the time from the EVAR procedure to hospital discharge or death.
Statistical analysis
Preoperative characteristics, medical risk factors, and procedural data were compared using univariate techniques, including χ2 or Fisher exact tests for categoric variables and univariate linear regression or Kruskal-Wallis tests for continuous variables. Characteristics are described using mean ± standard deviation or count (%).
Morbidity and mortality associations were examined using logistic regression. LOS was log-transformed before analysis with linear regression to satisfy normality assumptions. Pairwise LOS comparisons were back-transformed for presentation as percent differences.
All multivariable analyses were adjusted for age, race, sex, current smoking status, and total work relative value units of the component CPT codes defining the surgical procedure (to account for overall procedural complexity). Additional covariates were included in the analyses of each of the grouped morbidity classes as well as mortality. Covariates for the multivariable analyses were selected according to previous full-sample analyses of the ACS NSQIP by the central ACS NSQIP statistical faculty (which are available to each participating site) examining predictors of the morbidity classes detailed above and mortality in vascular surgery patients. The selected covariates for each analysis are detailed in the tabulated results that follow in this report. All analyses were performed using SAS 9.2 software (SAS Institute, Cary, NC).
RESULTS
Study sample characteristics
Demographic and risk factor data are summarized in Table I. We identified 6009 elective EVAR cases. Types of anesthesia administered were general in 4868 cases (81%), spinal in 419 (7%), epidural in 331 (5.5%), and local/MAC in 391 (6.5%). The study sample consisted of 5027 men (84%) and 982 women (16%), with a mean age of 74 years. Most patients (85%) were white.
Table I.
Patient characteristics by type of anesthesia during endovascular aneurysm repair (EVAR)
| Variablea | Elective EVAR (n = 6009) |
Anesthesia General (n = 4868) |
Spinal (n = 419) |
Epidural (n = 331) |
Local/MAC (n = 391) |
Pb |
|---|---|---|---|---|---|---|
| Age, years | 74.1 ± 8.5 | 73.8 ± 8.6 | 74.8 ± 7.7 | 74.8 ± 8.5 | 75.3 ± 8.9 | .0008 |
| Nonwhite race | 877 (14.6) | 695 (14.3) | 86 (20.5) | 32 (9.7) | 64 (16.4) | .0002 |
| Female sex | 982 (16.3) | 779 (16.0) | 71 (17.0) | 61 (18.4) | 71 (18.2) | .4716 |
| Body mass index, kg/m2 | 28.0 ± 5.6 | 28.1 ± 5.6 | 27.6 ± 6.9 | 28.3 ± 5.3 | 27.2 ± 4.9 | .0044 |
| Diabetes | 862 (14.4) | 707 (14.5) | 62 (14.8) | 38 (11.5) | 55 (14.1) | .4879 |
| Current smoker | 1690 (28.1) | 1400 (28.8) | 125 (29.8) | 74 (22.4) | 91 (23.3) | .0090 |
| Ever smoker | 3794 (77.5) | 3031 (77.1) | 289 (80.3) | 239 (82.4) | 235 (75.1) | .0704 |
| Functional status | ||||||
| Independent | 5747 (95.6) | 4668 (95.9) | 394 (94.0) | 318 (96.1) | 367 (93.9) | .0919 |
| Partially/totally dependent | 262 (4.4) | 200 (4.1) | 25 (6.0) | 13 (3.9) | 24 (6.1) | |
| History of | ||||||
| COPD | 1104 (18.4) | 885 (18.2) | 69 (16.5) | 65 (19.6) | 85 (21.7) | .2175 |
| Congestive heart failure | 82 (1.4) | 62 (1.3) | 3 (0.7) | 7 (2.1) | 10 (2.6) | .0680 |
| Myocardial infarction | 70 (1.2) | 65 (1.3) | 1 (0.2) | 0 | 4 (1.0) | .0227c |
| Angina | 132 (2.2) | 100 (2.1) | 9 (2.2) | 7 (2.1) | 16 (4.1) | .0715 |
| Prior CABG | 1461 (24.3) | 1186 (24.4) | 99 (23.6) | 88 (26.6) | 88 (22.5) | .6289 |
| Prior PTCI | 1265 (21.1) | 1047 (21.5) | 83 (19.8) | 49 (14.8) | 86 (22.0) | .0295 |
| Hypertension | 4758 (79.2) | 3868 (79.5) | 334 (79.7) | 246 (74.3) | 310 (79.3) | .1685 |
| Revascularization or amputation | 335 (5.6) | 261 (5.4) | 30 (7.2) | 15 (4.5) | 29 (7.4) | .1314 |
| Dialysis dependencec | 77 (1.3) | 58 (1.2) | 8 (1.9) | 5 (1.5) | 6 (1.5) | .5823 |
| Transient ischemic attack | 419 (7.0) | 335 (6.9) | 35 (8.4) | 22 (6.7) | 27 (6.9) | .7172 |
| Stroke | ||||||
| No residual disability | 280 (4.7) | 232 (4.8) | 19 (4.5) | 14 (4.2) | 15 (3.8) | .8320 |
| Residual disability | 299 (5.0) | 244 (5.0) | 22 (5.3) | 14 (4.2) | 19 (4.9) | .9229 |
| Transfer status | ||||||
| Other hospital or facility | 138 (2.3) | 107 (2.2) | 12 (2.9) | 9 (2.7) | 10 (2.6) | .7545 |
| Admitted directly from home | 5871 (97.7) | 4761 (97.8) | 407 (97.1) | 322 (97.3) | 381 (97.4) | |
| ASA class | ||||||
| 1–3 (no disturbance, mild, severe) | 4914 (81.8) | 3965 (81.5) | 365 (87.1) | 277 (83.7) | 307 (78.5) | .0078 |
| 4–5 (life-threatening, moribund) | 1093 (18.2) | 901 (18.5) | 54 (12.9) | 54 (16.3) | 84 (21.5) | |
| eGFR (mL/min)d | 68.6 ± 23.0 | 68.9 ± 23.0 | 68.2 ± 23.3 | 66.6 ± 23.0 | 67.8 ± 23.3 | .2683 |
| Chronic dyspnea | 1489 (24.8) | 1197 (24.6) | 93 (22.2) | 91 (27.5) | 108 (27.6) | .2038 |
ASA, American Society of Anesthesiologists; eGFR, estimated glomerular filtration rate; MAC, monitored anesthesia care; PTCI, percutaneous transluminal coronary intervention.
Data are shown as mean ± standard deviation or number (%).
P values for overall tests of differences among the four anesthesia categories from χ2 or Fisher exact test for categoric variables and one-way analysis of variance for continuous variables.
Fisher exact test used in place of χ2 (categoric). Dialysis indicated or creatinine level >6.0 mg/dL.
Set to 0 for preoperative dialysis.
Patient characteristics varied according to anesthesia type (Table I). Significant differences were observed according to anesthesia type received in age, race, prior history of myocardial infarction, prior history of percutaneous coronary revascularization, BMI, ASA class, and current smoking. The mean values and prevalence of these factors are summarized in Table I. Otherwise, comorbidities and risk factors were generally similar between the anesthesia groups.
Procedural specifics
Procedural specifics are summarized in Table II. The mean operative time for the study sample was 158 minutes, with the surgeon designated as a vascular surgeon in 98% of cases. A surgical resident was involved in 64% of cases. The mean transfusion requirement was 2.3 units in the 11% of patients who required a transfusion. Most cases used femoral artery access through a groin incision.
Table II.
Procedural specifics by type of anesthesia during endovascular aneurysm repair (EVAR)
| Variablea | Elective EVAR (n = 6009) |
Anesthesia General (n = 4868) |
Spinal (n = 419) |
Epidural (n = 331) |
Local/MAC (n = 391) |
Pb |
|---|---|---|---|---|---|---|
| Surgeon speciality | ||||||
| Vascular surgeon | 5868 (97.7) | 4738 (97.3) | 417 (99.5) | 330 (99.7) | 383 (98.0) | .0020 |
| Cardiothoracic or general surgeon | 141 (2.4) | 130 (2.7) | 2 (0.5) | 1 (0.3) | 8 (2.1) | |
| Resident involved | 3849 (64.2) | 3050 (62.8) | 259 (62.4) | 268 (81.5) | 272 (69.7) | <.0001 |
| Patients requiring transfusion | 666 (11.1) | 561 (11.5) | 24 (5.7) | 38 (11.5) | 43 (11.0) | .0042 |
| Units transfused (among those) | 2.3 ± 2.1 | 2.4 ± 2.2 | 1.7 ± 1.1 | 1.5 ± 0.7 | 1.7 ± 1.2 | <.0001c |
| Operative time (minutes) | 157.5 ± 67.2 | 159.7 ± 68.8 | 137.4 ± 53.1 | 156.5 ± 58.4 | 152.1 ± 63.4 | <.0001c |
| Access type | ||||||
| Percutaneousd | 2661 (44.3) | 2253 (46.3) | 159 (38.0) | 92 (27.8) | 157 (40.2) | <.0001 |
| Any open femoral access | 3348 (55.7) | 2615 (53.7) | 260 (62.1) | 239 (72.2) | 234 (59.9) | |
| Single iliac or aortic extension | 1666 (27.7) | 1305 (26.8) | 127 (30.3) | 93 (28.1) | 141 (36.1) | .0007 |
| Multiple extensions | 508 (8.5) | 396 (8.1) | 45 (10.7) | 26 (7.9) | 41 (10.5) | .1266 |
| Hypogastric embolization | 103 (1.7) | 88 (1.8) | 4 (1.0) | 8 (2.4) | 3 (0.8) | .1919 |
| Endograft configuration | ||||||
| Bifurcated modular | ||||||
| One docking limb | 2714 (45.2) | 2165 (44.5) | 156 (37.2) | 143 (43.2) | 250 (63.9) | <.0001 |
| Two docking limbs | 2158 (35.9) | 1741 (35.8) | 192 (45.8) | 144 (43.5) | 81 (20.7) | <.0001 |
| Unibody bifurcated | 441 (7.3) | 364 (7.5) | 26 (6.2) | 16 (4.8) | 35 (9.0) | .1399 |
| Aorto-uni-iliac | 257 (4.3) | 209 (4.3) | 19 (4.5) | 17 (5.1) | 12 (3.1) | .5598 |
| Aortoaortic tube | 480 (8.0) | 422 (8.7) | 28 (6.7) | 12 (3.6) | 18 (4.6) | .0003 |
| Total work relative value units | 38.7 ± 14.8 | 38.3 ± 14.8 | 41.2 ± 13.4 | 39.9 ± 12.8 | 40.8 ± 16.8 | <.0001 |
MAC, Monitored anesthesia care.
Continuous variables are expressed as mean ± standard deviation; categoric variables as number (%).
P values for overall tests of differences among the four anesthesia categories from χ2 test for categoric variables and one-way analysis of variance or Kruskal-Wallis test for continuous variables.
Kruskal-Wallis test used in place of one-way analysis of variance (continuous).
Defined by absence of a femoral code.
Procedural specifics varied according to anesthesia type (Table II). Significant differences were observed according to anesthesia type received in operative time, surgeon speciality, involvement of a surgical resident, and the need for transfusion.
Associations with morbidity, mortality, and LOS
Overall, defined morbidity occurred in 11% of patients, median LOS was 2 days (interquartile range, 1–3 days), and mean LOS was 2.8 ± 4.3 days. The 30-day mortality rate was 1.1%.
Rates of predefined end points are summarized in Table III. Significant univariable differences were observed in morbidity and LOS according to anesthesia type (Table IV). Univariable associations were observed between general anesthesia and an increase in any morbidity vs local/MAC (P = .018), pulmonary morbidity vs spinal (P = .010) and vs local/MAC (P = .042), and longer log-LOS vs spinal (P < .001) and vs local/MAC (P < .001).
Table III.
Outcomes by type of anesthesia during endovascular aneurysm repair (EVAR)
| Variablea | Elective EVAR (n = 6009) |
Anesthesia General (n = 4868) |
Spinal (n = 419) |
Epidural (n = 331) |
Local/MAC (n = 391) |
|---|---|---|---|---|---|
| Morbidity (any of below) | 677 (11.3) | 567 (11.7) | 37 (8.8) | 43 (13.0) | 30 (7.7) |
| Wound problems | |||||
| Superficial wound infection | 97 (1.6) | 84 (1.7) | 8 (1.9) | 4 (1.2) | 1 (0.3) |
| Deep wound infection | 34 (0.6) | 31 (0.6) | 2 (0.5) | 0 | 1 (0.3) |
| Organ space wound infection | 3 (0.1) | 1 (0.0) | 0 | 1 (0.3) | 1 (0.3) |
| Wound dehiscence | 16 (0.3) | 15 (0.3) | 0 | 1 (0.3) | 0 |
| Any superficial or deep wound infection | 130 (2.2) | 115 (2.4) | 9 (2.2) | 4 (1.2) | 2 (0.5) |
| Pulmonary | |||||
| Pneumonia | 73 (1.2) | 67 (1.4) | 2 (0.5) | 3 (0.9) | 1 (0.3) |
| Unplanned reintubation | 99 (1.7) | 91 (1.9) | 0 | 4 (1.2) | 4 (1.0) |
| Failure to wean from ventilator | 75 (1.3) | 71 (1.5) | 1 (0.2) | 2 (0.6) | 1 (0.3) |
| Any pulmonary morbidity | 168 (2.8) | 155 (3.2) | 3 (0.7) | 5 (1.5) | 5 (1.3) |
| Venous thromboembolic | |||||
| Deep venous thrombosis | 35 (0.6) | 30 (0.6) | 1 (0.2) | 3 (0.9) | 1 (0.3) |
| Pulmonary embolism | 9 (0.2) | 8 (0.2) | 0 | 1 (0.3) | 0 |
| Genitourinary | |||||
| Acute renal insufficiency | 35 (0.6) | 29 (0.6) | 2 (0.5) | 3 (0.9) | 1 (0.3) |
| Acute renal failure | 50 (0.8) | 46 (0.9) | 1 (0.2) | 2 (0.6) | 1 (0.3) |
| Urinary tract infection | 99 (1.7) | 81 (1.7) | 8 (1.9) | 5 (1.5) | 5 (1.3) |
| Any renal insufficiency or renal failure | 79 (1.3) | 69 (1.4) | 3 (0.7) | 5 (1.5) | 2 (0.5) |
| Cardiovascular | |||||
| Stroke | 27 (0.5) | 21 (0.4) | 2 (0.5) | 1 (0.3) | 3 (0.8) |
| Cardiac arrest | 25 (0.4) | 20 (0.4) | 1 (0.2) | 2 (0.6) | 2 (0.5) |
| Myocardial infarction | 15 (0.3) | 14 (0.3) | 0 | 1 (0.3) | 0 |
| Operative | |||||
| Postoperative hemorrhage/transfusions | 31 (0.5) | 26 (0.5) | 1 (0.2) | 3 (0.9) | 1 (0.3) |
| Graft failure | 56 (0.9) | 45 (0.9) | 2 (0.5) | 5 (1.5) | 4 (1.0) |
| Return to operating room | 260 (4.3) | 214 (4.4) | 13 (3.1) | 18 (5.4) | 15 (3.8) |
| Septic | |||||
| Sepsis | 62 (1.0) | 49 (1.0) | 6 (1.4) | 6 (1.8) | 1 (0.3) |
| Septic shock | 53 (0.9) | 46 (0.9) | 1 (0.2) | 3 (0.9) | 3 (0.8) |
| Death | 68 (1.1) | 60 (1.2) | 1 (0.2) | 4 (1.2) | 3 (0.8) |
| Post-op surgical length of stay, days | 2.8 ± 4.3 | 2.9 ± 4.2 | 2.1 ± 1.8 | 2.7 ± 5.1 | 2.3 ± 6.1 |
| Length of stay, days | 2 (1, 3) | 2 (1, 3) | 1 (1, 2) | 2 (1, 3) | 1 (1, 2) |
MAC, Monitored anesthesia care.
Data are presented as mean ± standard deviation, number (%), and median (interquartile range).
Table IV.
Regression model results for anesthesia groups vs morbidity and mortality outcomes
| Overall anesthesia |
General vs spinal | General vs epidural | General vs MAC/local | ||||
|---|---|---|---|---|---|---|---|
| Outcomes | P | OR (95% CI) | Pa | OR (95% CI) | Pa | OR (95% CI) | Pa |
| Renal morbidityb | |||||||
| Unadjusted | .3398 | ||||||
| Multivariable modelc | .2803 | ||||||
| Pulmonary morbidityb | |||||||
| Unadjusted | .0047 | 4.6 (1.5–14.4) | .0095 | 2.1 (0.9–5.3) | .0958 | 2.5 (1.0–6.2) | .0417 |
| Multivariable modeld | .0126 | 4.0 (1.3– 12.5) | .0195 | 1.9 (0.8–4.7) | .1711 | 2.6 (1.0–6.4) | .0409 |
| Mortalityb | |||||||
| Unadjusted | .3536 | ||||||
| Multivariable modele | .3463 | ||||||
| Wound morbidityb | |||||||
| Unadjusted | .0958 | ||||||
| Multivariable modelf | .1403 | ||||||
| Any morbidityb | |||||||
| Unadjusted | .0279 | 1.4 (1.0–1.9) | .0831 | 0.9 (0.6–1.2) | .4626 | 1.6 (1.1–2.3) | .0181 |
| Multivariable modelg | .0514 | ||||||
| % Difference (95% CI) | P | % Difference (95% CI) | P | % Difference (95% CI) | P | ||
| Length of stayh | |||||||
| Unadjusted | <.0001 | 16.80 (10.8–23.1) | <.0001 | 5.20 (–0.9 to 11.6) | .0945 | 20.10 (13.7–26.9) | <.0001 |
| Multivariable modelg | <.0001 | 10.00 (4.8–15.5) | .0001 | 3.60 (–1.9 to 9.4) | .2046 | 20.00 (14.1–26.2) | <.0001 |
CI, Confidence interval; MAC, monitored anesthesia care; OR, odds ratio.
P value for contrast.
Logistic regression.
Model covariates for renal: age, gender, race, current smoker, total work relative value units, femoral access, American Society of Anesthesiologists class, estimated glomerular filtration rate, functional status, body mass index, diabetes, hypertension, history of congestive heart failure.
Model covariates for pulmonary: age, gender, race, current smoker, total work relative value units, femoral access, American Society of Anesthesiologists class, estimated glomerular filtration rate, functional status, history of congestive heart failure, chronic obstructive pulmonary disease, angina, cerebrovascular accident; red blood cell transfusions, dyspnea.
Model covariates for mortality: age, gender, race, current smoker, total work relative value units, femoral access, American Society of Anesthesiologists class, estimated glomerular filtration rate functional status, body mass index; history of congestive heart failure, chronic obstructive pulmonary disease, angina, cerebrovascular accident; red blood cell transfusions, prior percutaneous transluminal coronary intervention, prior coronary artery bypass grafting.
Model covariates for wound: age, gender, race, current smoker, total work relative value units, femoral access, body mass index, diabetes, history of chronic obstructive pulmonary disease, operative time, long-term steroid use.
Model covariates for any morbidity and length of stay: age, gender, race, current smoker, total work relative value units, femoral access, American Society of Anesthesiologists class, estimated glomerular filtration rate, functional status, body mass index, diabetes; history of congestive heart failure, chronic obstructive pulmonary disease, angina, cerebrovascular accident; red blood cell transfusions, operative time.
Linear regression (log-transformed for analysis, back-transformed results are presented).
Multivariable analyses of morbidity, mortality, and LOS are summarized in Table IV. Significant multivariable differences were observed in morbidity and LOS according to anesthesia type. Use of general anesthesia was associated with a significant increase in pulmonary morbidity compared with spinal (odds ratio [OR], 4.0; 95% confidence interval [CI], 1.3–12.5; P = .020) and local/MAC anesthesia (OR, 2.6; 95% CI, 1.0–6.4; P = .041). Other significant predictors of increased pulmonary morbidity included current smoking, lower eGFR, ASA class 4 or 5, partial or total functional dependence, chronic obstructive pulmonary disease (COPD), and the volume of any necessary transfusion. Complete multivariable model results for pulmonary morbidity are presented in Table V (Hosmer-Lemeshow goodness of fit; P > .05).
Table V.
Complete multivariable logistic regression results for pulmonary morbidity
| Variable | OR (95% CI) | P |
|---|---|---|
| General anesthesia vs | .0126 | |
| Spinal | 3.96 (1.25–12.54) | .0195 |
| Epidural | 1.89 (0.76–4.73) | .1711 |
| Monitored anesthesia care/local | 2.58 (1.04–6.41) | .0409 |
| Age, years | 1.01 (0.99–1.03) | .5297 |
| Female | 1.14 (0.77–1.69) | .5139 |
| Nonwhite | 1.47 (0.98–2.21) | .0663 |
| Current smoker | 1.55 (1.07–2.23) | .0197 |
| Total work relative value units | 1.00 (0.99–1.01) | .9202 |
| Access (percutaneous vs femoral) | 1.48 (1.0–2.18) | .0504 |
| ASA class (4–5 vs 1–3) | 1.50 (1.04–2.16) | .0287 |
| Estimated glomerular filtration rate | 0.99 (0.98–0.99) | .0004 |
| Functionally dependent (partial/total vs independent) | 2.09 (1.23–3.56) | .0064 |
| History of | ||
| Congestive heart failure | 1.93 (0.85–4.40) | .1164 |
| Congestive obstructive pulmonary disease | 1.71 (1.16–2.53) | .0069 |
| Angina | 1.35 (0.56–3.28) | .5050 |
| Cerebrovascular accident | 0.92 (0.56–1.52) | .7394 |
| Units transfused | 1.32 (1.22–1.43) | <.0001 |
| Chronic dyspnea | 1.13 (0.77–1.65) | .5416 |
ASA, American Society of Anesthesiologists; CI, confidence interval; OR, odds ratio.
Use of general anesthesia was also significantly associated with a prolonged LOS, with a 10% increase for general anesthesia compared with spinal (95% CI, 4.8%–15.5%; P = .001) and a 20% increase for general anesthesia compared with local/MAC (95% CI, 14.1%–26.2%; P < .001). Other significant predictors of increased LOS included age, female sex, nonwhite race, ASA class 4 or 5, percutaneous femoral artery access, decreased eGFR, any level of functional dependence other than independence; a history of congestive heart failure, COPD, angina, or diabetes; increased volume of necessary transfusion, and operative time. Complete multivariable model results for LOS are presented in Table VI.
Table VI.
Multivariable linear regression results for log-length of stay
| Variable | β (95% CI) | P |
|---|---|---|
| General anesthesia vs | <.0001 | |
| Spinal | 0.095 (0.046–0.144) | .0001 |
| Epidural | 0.035 (−0.019 to 0.090) | .2046 |
| Monitored anesthesia care/local | 0.182 (0.132–0.232) | <.0001 |
| Age, years | 0.003 (0.002–0.005) | <.0001 |
| Female | 0.101 (0.067–0.136) | <.0001 |
| Nonwhite | 0.075 (0.040–0.111) | <.0001 |
| Current smoker | 0.008 (−0.023 to 0.037) | .6608 |
| Total work relative value units | 0.000 (−0.001 to 0.001) | .7934 |
| Access (percutaneous vs femoral) | 0.030 (0.000 0.060) | .0495 |
| ASA class (4–5 vs 1–3) | 0.100 (0.068–0.133) | <.0001 |
| Estimated glomerular filtration rate | −0.002 (−0.002 to −0.001) | <.0001 |
| Functionally dependent (partial/total vs independent) | 0.285 (0.223–0.347) | <.0001 |
| Body mass index | −0.001 (−0.004 to 0.001) | .2725 |
| History of | ||
| Congestive heart failure | 0.294 (0.187–0.402) | <.0001 |
| Diabetes | 0.040 (0.005–0.076) | .0259 |
| Congestive obstructive pulmonary disease | 0.072 (0.040–0.105) | <.0001 |
| Angina | 0.121 (0.037–0.205) | .0048 |
| Cerebrovascular accident | 0.037 (−0.005 to 0.079) | .0867 |
| Units transfused | 0.085 (0.071–0.099) | <.0001 |
| Operative time, min | 0.002 (0.0017–0.0021) | <.0001 |
ASA, American Society of Anesthesiologists; CI, confidence interval.
Significant differences relative to general anesthesia were not observed for epidural anesthesia use with regard to decreased pulmonary morbidity or LOS.
DISCUSSION
The anesthetic techniques used during EVAR varied widely across North America, with general anesthesia being the most common, by a large margin. Although this was an observational study, the data presented suggest significant advantages to the use of spinal and local/MAC anesthesia compared with general anesthesia in the performance of elective EVAR. In the examined sample of 6009 elective EVAR cases from 211 North American hospitals from 2005 to 2008, the use of general anesthesia was associated with higher postoperative pulmonary complications and LOS than spinal and local/MAC anesthesia, even when adjusting for other important patient characteristics.
EVAR was developed as a less-invasive and potentially safer alternative to traditional open surgical repair for aortic aneurysms.1 The most common surgical method for introduction of the EVAR device is a groin incision, and no aortic cross-clamping is required. As such, the procedure lends itself to local and regional anesthesia techniques in a way that open aneurysm repair does not. However, surgical teams have yet to discern in which patients this advantage can be most effectively used. Early reports from the investigational and European use of EVAR demonstrated the feasibility of nongeneral anesthesia and suggested locoregional anesthesia might have benefits.
The first report describing the feasibility of local anesthesia for EVAR was published by Henretta et al6 in 1999. That report detailed no deaths or significant morbidity in a series of 47 consecutive patients. Also in 1999, Cao et al5 reported results in 115 patients undergoing EVAR, with 61 receiving epidural anesthesia. Epidural anesthesia was associated with a reduction in the total hospital LOS and a lower utilization of the intensive care unit (ICU) compared with general anesthesia.
The era of widespread American use of EVAR began in 1999 after U.S. Food and Drug Administration (FDA) approval of the AneuRx (Medtronic, Minneapolis, Minn) and the Guidant EVT (Guidant Corp, Indianapolis, Ind) devices for use. Since that time, other single-center reports of EVAR using locoregional anesthesia techniques have been published. In 2002, de Virgilio et al7 published a report demonstrating equivalent safety for local and general anesthesia. However, their results demonstrated no differences in mortality, cardiac events, or pulmonary events in an examination of 229 patients undergoing EVAR during a 4-year period. In 2005, Verhoeven et al11 reported that the use of primary local anesthesia was safe and associated with improvements in pulmonary morbidity and decreased ICU stay.
These single-center, retrospective reports demonstrated the safety of locoregional anesthetic techniques for EVAR compared with general anesthesia; furthermore, all but one suggested that significant benefits might be associated with these techniques. The studies were limited, however, because they reflected single-center practice patterns and included a relatively small number of patients in their analyses.
Larger multicenter observational studies of these issues have also been reported. Parra et al8 reported associations among EVAR performed under local anesthesia with decreased LOS and a decrease in a variety of morbid events (including renal, wound, and cardiac complications) in a retrospective analysis of 424 patients undergoing EVAR as part of a phase II trial of the AneuRx device (Medtronic). More recently, investigators have used the European Collaborators on Stent-Graft Techniques for Aortic Aneurysm Repair (EUROSTAR) registry of EVAR procedures to examine issues of anesthesia type and outcome. In an initial analysis of 5557 patients undergoing EVAR, Ruppert et al9 reported decreased ICU admissions, LOS, and cardiac complications in multivariable analyses of locoregional anesthesia techniques. In that report, the greatest reductions were seen with local anesthesia techniques for EVAR.
The same group reported additional analyses of the EUROSTAR data in 2007 with patients stratified into high-risk and low-risk categories according to ASA class.10 In that stratified analysis, the most significant benefits from local and regional anesthesia were seen in patients designated as high risk. Decreased ICU utilization was observed with local anesthesia and decreased early mortality was observed with regional anesthesia in high-risk patients. Other than this latter analysis of high-risk patients, none of these studies have demonstrated any effect on mortality with differing anesthetic techniques.
Our observations are consistent with many of these previously referenced findings. Our analysis showed the use of local/MAC and spinal anesthesia was associated with a significant decrease in pulmonary complications compared with general anesthesia in multivariable models adjusting for pertinent risk factors. Further, a decreased LOS was also seen favoring the use of local/MAC and spinal anesthesia. Significant associations relative to general anesthesia were not observed for epidural anesthesia. These findings were observed in a large sample of EVAR cases sampled from contemporary medical practice in North American hospitals and provide a “real-world” description of both the use of less-invasive anesthetic techniques and the relative value of those techniques in limiting morbidity.
The EUROSTAR registry findings are quite similar to those reported here. Health systems in Europe, however, possess features that are quite different from the system in the United States, which may limit the applicability of their findings to American medical practice. In the initial EUROSTAR report by Ruppert et al,9 the average LOS was 5.8 days, which is approximately double the LOS for EVAR observed in this report. This likely represents a greater caution toward discharge in the early days of EVAR, because EUROSTAR contains data from 1997 to 2004, and the ACS NSQIP contains data from 2005 to 2008. In addition, the devices available for use in European countries also differ from those approved by the FDA. Furthermore, the relative availability of EVAR to the general patient population in Europe relative to North America may differ because of differences in the respective health care delivery systems, which limit (and in some cases prohibit) EVAR or other surgical procedures in high-risk or elderly European patients. To this point, the high-risk patients described in Ruppert et al10 described all patients with ASA class ≥3 as high risk, which would have included 93% of the study sample for this report.
European biases toward a more aggressive use of less-invasive anesthesia also likely play a role in the differences in outcomes among European and American studies. These differences are demonstrated in the distributions of general, regional, and local anesthesia reported in EUROSTAR (69%, 25%, and 6%) and the ACS NSQIP (81%, 13%, and 6%). The sum of these differences is significant. As such, this study represents the first large-scale demonstration of locoregional anesthetic benefit from American sources using contemporary technology.
As a separate concern, most of the preceding data presented regarding the benefits of using regional anesthesia have been derived from the use of epidural anesthesia. This report demonstrates clear benefits from the use of local/MAC and spinal anesthesia, but not epidural anesthesia, especially in relation to LOS. This finding is clearly different from the results of the preceding single-center and multicenter data presented and represents a unique and intriguing finding. The observed lack of significant associations between epidural anesthesia and decreased pulmonary morbidity or LOS could have been due to the effects of local anesthetic or adjuvant agents commonly used within epidural anesthetics. The volume and concentration of local anesthetics affects epidural dermatomal spread as well as the depth of analgesia or the degree of motor blockade, or both. When patients are supine, cephalad spread of local anesthetics or epidural narcotics to the midthoracic or lower cervical regions during a continued infusion may impair pulmonary mechanics, thus increasing the risk of postoperative pulmonary dysfunction and LOS. Narcotic adjuvants are also frequently added to local anesthetic epidural infusions. These agents also have the potential to centrally suppress respiratory drive and thereby affect rates of reintubation and LOS.
Potential mechanisms by which locoregional anesthesia may affect the morbidity of EVAR include avoidance of endotracheal intubation and mechanical ventilation as well as the potential for residual neuromuscular paralysis after reversal. Atelectasis ensues immediately after induction of general anesthesia and persists well into the postoperative phase.16,17 Positive pressure ventilation results in atelectasis, diminishes functional residual capacity, and increases the risk of reintubation, especially in the elderly.18 High-risk patients (ASA 4 and 5 and elderly) are at a greater risk of postoperative pulmonary morbidity with general anesthesia, even after what appears to be adequate reversal of neuromuscular blockade. Furthermore, general anesthesia frequently involves the use of volatile anesthetics during the maintenance phase. Inhalational agents have been associated with immunosuppression, potentially increasing the risk of postoperative pneumonia in the setting of atelectasis, diminished cough reflex, or in patients at risk for aspiration.19,20
Locoregional anesthesia avoids mechanical ventilation and permits maintenance of spontaneous ventilation, thereby minimizing the patient’s exposure to factors that increase the risk of postoperative pulmonary failure. Furthermore, locoregional anesthetic techniques provide preemptive analgesia and improve postoperative pain control relative to general anesthesia alone, which may reduce the incidence of hypertension and tachycardia related to surgical stress and postoperative pain. Locoregional techniques also allow for avoidance of ventilator weaning at the end of anesthesia, which can be challenging in compromised patients.
Our study showed that local/MAC and spinal anesthesia techniques were able to reduce LOS by 10% to 20%, even in the context of the shorter observed American hospital LOS. These anesthetic techniques were also associated with a 60% to 75% decrease in the odds of postoperative pulmonary complications, including pneumonia and failure to wean from the ventilator ≤48 hours of surgery. These complications obviously increase patient discomfort and commonly require admission to the ICU or extension of the ICU stay. Given the high estimated cost of such nosocomial pneumonias (>$12,000 per occurrence21) and the potential savings of the observed decreases in LOS, the significance of these data to contemporary American health care is obvious.
An obvious explanation, however, is not apparent for the differences in spinal and epidural anesthesia in their observed advantages relative to general anesthesia in the performance of EVAR. It is possible that the smaller sample receiving epidural anesthesia resulted in the lack of a significant association, which would be suggested by the observed trends toward an association (especially with regard to pulmonary complications). Plausible mechanisms can be put forth, though, to explain the observed differences. In addition to the potential mechanisms underlying the lack of an observed association discussed above, it is also possible that epidural anesthesia was associated with increased crystalloid fluid administration due to prolonged sympatholysis relative to spinal anesthesia and that this fluid volume was associated with a slightly higher rate of pulmonary complication (and accordingly LOS). It is also possible that epidural anesthetics were associated with a small but significant incidence of complications found with the more complicated technique involving catheter placement (ie, postepidural headache), which may also have resulted in occasional increases in LOS. Unfortunately, the data points collected as part of NSQIP will not allow these questions to be addressed.
This study possesses a number of other limitations that deserve comment. As with almost all secondary analyses of existing databases, the database was not specifically designed to assess the question of interest. The ACS NSQIP is a quality-of-care tool that is not optimized to assess specifics of endovascular surgical and anesthetic care. Detailed data regarding patient anatomy, device selection, anesthetic techniques, monitoring, or conversion between anesthetic techniques were not collected and cannot be analyzed.
Patients were also not randomized to their anesthesia type, eliminating any control for known (or unknown) confounding factors. Individual anesthetic selections were made according to surgeon and anesthesiologist preference, incorporating the biases of those individuals as well as patient-specific factors such as anatomy, medical risk, available resources, and procedural complexity. As such, this report represents a large, uncontrolled investigation of anesthetic technique as it was applied in contemporary medical practice and was not a test of an a priori hypothesis.
Furthermore, data were also lacking for analysis regarding surgeon and institutional identity. This raises the possibility that the biases of individuals and institutions and the potential for unbalanced confounding factors in the absence of randomization, such as the observed tendency for general anesthesia for longer (and presumably more complex) cases, may have produced the observed differences. The likelihood, though, of such a type I error is mitigated somewhat by the use of multivariable methods to control for potential confounding factors as identified by univariable analyses of these data and prior analyses of morbidity and mortality in vascular surgery patients throughout the ACS NSQIP.
It is also possible that other missing data points such as the time of surgery, the day of the week of surgery, and urinary retention affected our findings, especially with regard to LOS. Further mitigation of unmeasured confounding would require more detailed data or advanced observational methods, such as propensity score modeling, hierarchical modeling, or instrumental variable analysis. In the analysis we have described, implementation of these methods is limited by the relatively small number of patients in each exposure category, as well as by the relatively small proportion of high-risk pulmonary patients.
CONCLUSIONS
This study represents a large-sample investigation of the benefits and liabilities of differing anesthesia types in the performance of EVAR in North America using contemporary anesthetic and surgical techniques. The results suggest that the use of general anesthesia for the performance of EVAR is associated with higher rates of pulmonary morbidity and a 10% to 20% increase in LOS relative to locoregional anesthetic techniques, specifically local/MAC and spinal anesthesia. These data support an increase in the use of local anesthesia/MAC or spinal anesthesia in EVAR patients suitable for such anesthetic approaches to reduce pulmonary morbidity and length of stay.
Footnotes
Competition of interest: none.
Presented at the 2010 Annual Meeting of the American Society of Anesthesiologists, San Diego, Calif, Oct 16–20, 2010.
AUTHOR CONTRIBUTIONS
Conception and design: ME, AE, MC, PG
Analysis and interpretation: ME, JS, RG, MC, PG, CG,KH
Data collection: Not applicable
Writing the article: ME, JS, AE, RG
Critical revision of the article: ME, JS, AE, RG, MC, PG, CG, KH
Final approval of the article: ME, JS, AE, RG, MC, PG, CG, KH
Statistical analysis: JS
Obtained funding: Not applicable
Overall responsibility: ME
REFERENCES
- 1.Parodi JC, Palmaz JC, Barone HD. Transfemoral intraluminal graft implantation for abdominal aortic aneurysms. Ann Vasc Surg. 1991;5:491–499. doi: 10.1007/BF02015271. [DOI] [PubMed] [Google Scholar]
- 2.Blankensteijn JD, de Jong SE, Prinssen M, van der Ham AC, Buth J, et al. Two-year outcomes after conventional or endovascular repair of abdominal aortic aneurysms. N Engl J Med. 2005;352:2398–2405. doi: 10.1056/NEJMoa051255. [DOI] [PubMed] [Google Scholar]
- 3.De Bruin JL, Baas AF, Buth J, Prinssen M, Verhoeven EL, Cuypers PW, et al. Long-term outcome of open or endovascular repair of abdominal aortic aneurysm. N Engl J Med. 2010;362:1881–1889. doi: 10.1056/NEJMoa0909499. [DOI] [PubMed] [Google Scholar]
- 4.de Virgilio C, Bui H, Donayre C, Ephraim L, Lewis RJ, Elbassir M, et al. Endovascular vs open abdominal aortic aneurysm repair – A comparison of cardiac morbidity and mortality. Arch Surg. 1999;134:947–950. doi: 10.1001/archsurg.134.9.947. [DOI] [PubMed] [Google Scholar]
- 5.Cao P, Zannetti S, Parlani G, Verzini F, Caporali S, Spaccatini A, et al. Epidural anesthesia reduces length of hospitalization after endoluminal abdominal aortic aneurysm repair. J Vasc Surg. 1999;30:651–657. doi: 10.1016/s0741-5214(99)70104-7. [DOI] [PubMed] [Google Scholar]
- 6.Henretta JP, Hodgson KJ, Mattos MA, Karch LA, Hurlbert SN, Sternbach Y, et al. Feasibility of endovascular repair of abdominal aortic aneurysms with local anesthesia with intravenous sedation. J Vasc Surg. 1999;29:793–798. doi: 10.1016/s0741-5214(99)70205-3. [DOI] [PubMed] [Google Scholar]
- 7.de Virgilio C, Romero L, Donayre C, Meek K, Lewis RJ, Lippmann M, et al. Endovascular abdominal aortic aneurysm repair with general versus local anesthesia: a comparison of cardiopulmonary morbidity and mortality rates. J Vasc Surg. 2002;36:988–991. doi: 10.1067/mva.2002.128314. [DOI] [PubMed] [Google Scholar]
- 8.Parra JR, Crabtree T, McLafferty RB, Ayerdi J, Gruneiro LA, Ramsey DE, et al. Anesthesia technique and outcomes of endovascular aneurysm repair. Ann Vasc Surg. 2005;19:123–129. doi: 10.1007/s10016-004-0138-y. [DOI] [PubMed] [Google Scholar]
- 9.Ruppert V, Leurs LJ, Steckmeier B, Buth J, Umscheid T. Influence of anesthesia type on outcome after endovascular aortic aneurysm repair: an analysis based on Eurostar data. J Vasc Surg. 2006;44:16–21. doi: 10.1016/j.jvs.2006.03.039. [DOI] [PubMed] [Google Scholar]
- 10.Ruppert V, Leurs LJ, Rieger J, Steckmeier B, Buth J, Umscheid T, et al. Risk-adapted outcome after endovascular aortic aneurysm repair: analysis of anesthesia types based on Eurostar data. J Endovasc Ther. 2007;14:12–22. doi: 10.1583/06-1957.1. [DOI] [PubMed] [Google Scholar]
- 11.Verhoeven ELG, Cina CS, Tielliu IFJ, Zeebregts CJ, Prins TR, Eindhoven GB, et al. Local anesthesia for endovascular abdominal aortic aneurysm repair. J Vasc Surg. 2005;42:402–409. doi: 10.1016/j.jvs.2005.05.047. [DOI] [PubMed] [Google Scholar]
- 12.Crawford RS, Cambria RP, Abularrage CJ, Conrad MF, Lancaster RT, Watkins MT, et al. Preoperative functional status predicts perioperative outcomes after infrainguinal bypass surgery. J Vasc Surg. 2010;51:351–359. doi: 10.1016/j.jvs.2009.08.065. [DOI] [PubMed] [Google Scholar]
- 13.Hua HT, Cambria RP, Chuang SK, Stoner MC, Kwolek CJ, Rowell KS, et al. Early outcomes of endovascular versus open abdominal aortic aneurysm repair in the National Surgical Quality Improvement Program-Private Sector (NSQIP-PS) J Vasc Surg. 2005;41:382–389. doi: 10.1016/j.jvs.2004.12.048. [DOI] [PubMed] [Google Scholar]
- 14.LaMuraglia GM, Conrad MF, Chung T, Hutter M, Watkins MT, Cambria RP. Significant perioperative morbidity accompanies contemporary infrainguinal bypass surgery: an NSQIP report. J Vasc Surg. 2009;50:299–304. doi: 10.1016/j.jvs.2009.01.043. [DOI] [PubMed] [Google Scholar]
- 15.Levey AS, Coresh J, Greene T, Stevens LA, Zhang YP, Hendriksen S, et al. Using standardized serum creatinine values in the modification of diet in renal disease study equation for estimating glomerular filtration rate. Ann Intern Med. 2006;145:247–254. doi: 10.7326/0003-4819-145-4-200608150-00004. [DOI] [PubMed] [Google Scholar]
- 16.Edmark L, Kostova-Aherdan K, Enlund M, Hedenstierna G. Optimal oxygen concentration during induction of general anesthesia. Anesthesiology. 2003;98:28–33. doi: 10.1097/00000542-200301000-00008. [DOI] [PubMed] [Google Scholar]
- 17.Gunnarsson L, Tokics L, Gustavsson H, Hedenstierna G. Influence of age on atelectasis formation and gas exchange impairment during general anaesthesia. Br J Anaes. 1991;66:423–432. doi: 10.1093/bja/66.4.423. [DOI] [PubMed] [Google Scholar]
- 18.Weingarten TN, Whalen FX, Warner DO, Gajic O, Schears GJ, Snyder MR, et al. Comparison of two ventilatory strategies in elderly patients undergoing major abdominal surgery. Br J Anaes. 2010;104:16–22. doi: 10.1093/bja/aep319. [DOI] [PubMed] [Google Scholar]
- 19.Brand JM, Kirchner H, Poppe C, Schmucker P. The effects of general anesthesia on human peripheral immune cell distribution and cytokine production. Clin Immuno Immunopathol. 1997;83:190–194. doi: 10.1006/clin.1997.4351. [DOI] [PubMed] [Google Scholar]
- 20.Chang CC, Lin HC, Lin HW, Lin HC. Anesthetic management and surgical site infections in total hip or knee replacement: a population-based study. Anesthesiology. 2010;113:279–284. doi: 10.1097/ALN.0b013e3181e2c1c3. [DOI] [PubMed] [Google Scholar]
- 21.Warren DK, Shukla SJ, Olsen MA, Kollef MH, Hollenbeak CS, Cox MJ, et al. Outcome and attributable cost of ventilator-associated pneumonia among intensive care unit patients in a suburban medical center. Crit Care Med. 2003;31:1312–1317. doi: 10.1097/01.CCM.0000063087.93157.06. [DOI] [PubMed] [Google Scholar]