Intraoperative Fluid Therapy and Pulmonary Complications
Total Page:16
File Type:pdf, Size:1020Kb
■ Feature Article Intraoperative Fluid Therapy and Pulmonary Complications KRZYSZTOF SIEMIONOW, MD; JACEK CYWINSKI, MD; KRZYSZTOF KUSZA, MD, PHD; ISADOR LIEBERMAN, MD, MBA, FRCSC abstract Full article available online at ORTHOSuperSite.com. Search: 20120123-06 The purpose of this study was to evaluate the effects of intraoperative fl uid therapy on length of hospital stay and pulmonary complications in patients undergoing spine surgery. A total of 1307 patients were analyzed. Sixteen pulmonary complications were observed. Patients with a higher volume of administered crystalloids, colloids, and total intravenous fl uids were more likely to have postoperative respiratory com- plications: the odds of postoperative respiratory complications increased by 30% with an increase of 1000 mL of crystalloid administered. The best cutoff point for total fl uids was 4165 mL, with a sensitivity of 0.8125 and specifi city of 0.7171, for postoperative pulmonary complications. A direct correlation existed between fl uids and length of stay: patients who received Ͼ4165 mL of total fl uids had an average length of stay of 3.88Ϯ4.66 days vs 2.3Ϯ3.9 days for patients who received Ͻ4165 mL of total fl uids (PϽ.0001). This study should be considered as hypothesis-generating to design a prospective trial comparing high vs low intraoperative fl uid regiments for patients undergoing spine surgery. Dr Siemionow is from the Department of Orthopaedic Surgery, University of Illinois, Chicago, Illinois; Dr Cywinski is from the Department of Anesthesia, Cleveland Clinic, Cleveland, Ohio; Dr Kusza is from the Department of Anesthesia, Centrum Medyczne Bydgoszcz, Bydgoszcz, Poland; and Dr Lieberman is from Texas Back Institute, Plano, Texas. Drs Siemionow, Cywinski, Kusza, and Lieberman have no relevant fi nancial relationships to disclose. Correspondence should be addressed to: Krzysztof Siemionow, MD, Department of Orthopaedic Surgery, University of Illinois, 835 S Wolcott Ave, Room E-270, Chicago, IL 60612 ([email protected]). doi: 10.3928/01477447-20120123-06 e184 ORTHOPEDICS | ORTHOSuperSite.com INTRAOPERATIVE FLUID THERAPY | SIEMIONOW ET AL ntraoperative fl uid administration open abdominal aortic aneurysm repair. This included patients who underwent strategies vary widely as a result of In a study of 1280 consecutive patients vertebral augmentation procedures, mi- Ithe limited understanding of the phys- undergoing isolated on-pump coronary ar- crodiskectomies, single and multilevel de- iologic and clinical consequences of dif- tery bypass grafting, a total fl uid balance compressive procedures, single and mul- ferent fl uid replacement regimens on post- Ͼ500 mL at the end of the surgical pro- tilevel fusions, and osteotomies. Patients operative outcomes.1 The stress response cedure was predictive of increased length who did not have a reported intraoperative to surgery has been shown to profoundly of stay.7 estimated blood loss (EBL) or total vol- alter fl uid homeostasis, leading to fl uid In patients undergoing elective spine ume of crystalloids, colloids, red blood conservation, maintenance of intravas- surgery, intraoperative blood loss var- cells (RBC), fresh frozen plasma (FFP), cular volume, and preservation of blood ies but can be signifi cant (1 to 2 L).8 platelets administered, or urine output perfusion to vital organs. Primary media- Intraoperative blood loss is replaced with were excluded from the analysis. tors of surgical stress response are the re- various combinations of crystalloids, col- Respiratory complications were de- nin–angiotensin system, aldosterone, and loids, and blood products; however, the fi ned as clinical diagnosis of postopera- antidiuretic hormone; however, the effect replacement strategy varies among clini- tive acute respiratory distress syndrome, of the different fl uid regimens (liberal vs cians, with no clear guideline as to the op- clinical diagnosis of pulmonary edema, restrictive) on these mediators is largely timal management of intraoperative fl uid or clinical diagnosis of respiratory insuffi - unknown. Furthermore, the infl amma- replacement. Patients undergoing spinal ciency. The diagnosis was made by a con- tory response to surgical stress results in procedures differ signifi cantly from pa- sulting service, such as internal medicine, increased capillary permeability and a tients undergoing abdominal or thoracic pulmonology, or cardiology. Patients with tendency toward edema formation.2 Thus, surgery. Patients undergoing spine sur- a diagnosis of a pulmonary embolus and controversy exists about the optimal fl uid gery are often in the prone position, un- deep vein thrombosis were excluded from replacement strategy and the consequenc- dergo surgeries of varying time lengths, this study. Patients with obvious nonanes- es of intraoperative fl uid management on require anterior exposures through the thetic causes (eg, postoperative seizure or postoperative outcomes. chest or abdomen, and undergo signifi cant aspiration pneumonia) were also excluded Intraoperative administration of 40 bone and soft tissue manipulation, curve from this study. mL/kg compared with 15 mL/kg of lac- correction, osteotomy, or neural manipu- All intraoperative events, including tated Ringer’s solution was shown to lation, all of which can result in signifi - electrocardiogram changes (eg, tachy- improve postoperative organ functions cant blood loss. cardia, bradycardia, or arrythmias), hy- and recovery and shorten the length of The purpose of this retrospective study potension (Ͻ90 systolic blood pressure), stay in patients undergoing laparoscopic was to evaluate the effects of intraop- decreased partial pressure of oxygen cholecystectomy.3 However, the same in- erative fl uid therapy on length of hospital (Ͻ90%), fl uid balance (fl uids in minus vestigators later reported that restrictive stay and pulmonary complications in pa- fl uid outs), and American Society of (median, 1640 mL; range, 935-2250 mL) tients undergoing spine surgery. Anesthesiologists (ASA) scores, were ob- compared with liberal fl uid administra- tained from the intraoperative anesthesia tion (median, 5050 mL; range, 3563-8050 MATERIALS AND METHODS record. mL) led to a signifi cant improvement in After Institutional Review Board ap- Categorical variables were sum- pulmonary function and a reduction of proval, a review of the Cleveland Clinic marized using frequency and percent. postoperative hypoxemia.4 Department of General Anesthesiology Continuous variables were summarized Functional residual capacity decreases Perioperative Health Documentation using mean and standard deviation. Chi- by 10% and diffusing capacity by 6% in System was undertaken. The Perioperative square test or Fisher’s exact test was used response to an intravenous (IV) saline Health Documentation System combines to compare categorical variables between load of 22 mL/kg in healthy volunteers.5 pre-, intra-, and postoperative in-hospital patients with and without postoperative Neither parameter returned to normal up and post-discharge follow-up health in- respiratory complications. A 2-sample to 40 minutes after the infusion. McArdle formation fed automatically by compo- t test was used to compare continuous et al6 reported that a positive fl uid balance nents of the hospital electronic medical variables. A univariate logistic regression is predictive of major adverse morbid record, anesthesia information manage- model was used to analyze the associa- events (eg, myocardial infarction, pulmo- ment system, administrative claims data, tion between the crystalloid, colloid, and nary edema), increased intensive care unit and provider-entered health information. total fl uids and postoperative respiratory stay, and increased overall length of stay All patients who underwent spine surgery complications. Receiver operating char- in 100 consecutive patients undergoing between 2005 and 2007 were included. acteristic (ROC) analysis predicting post- FEBRUARY 2012 | Volume 35 • Number 2 e185 ■ Feature Article Table 1 Categorical Variables in Complication Versus No Complication Groups operative respiratory complications was performed based on crystalloid, colloid, Complication No Complication and total fl uids, respectively. Sensitivity Variable Total Group, N (%) Group, N (%) P a and specifi city for each predictor was cal- Sex .19 Female 604 10 (62.5) 594 (46) culated. Sensitivity is the predicted proba- Male 703 6 (37.5) 697 (54) bility of postoperative respiratory compli- Race .083b cation when a patient has a complication. Other 140 4 (25) 136 (10.5) Specifi city is the predicted probability of White 1167 12 (75) 1155 (89.5) no complication when a patient has no Preop CAD .62b complication. The best cutoff point for No 1228 16 (100) 1212 (93.9) each predictor was chosen by minimizing Yes 79 0 (0) 79 (6.1) the distance to the perfect point with sen- Preop MI .99b sitivity of 1 and specifi city of 1. Based on No 1304 16 (100) 1288 (99.8) the total fl uid cutoff point, a subanalysis Yes 3 0 (0) 3 (0.23) b was performed to correlate this predictor Preop hyperlipidemia .15 No 1118 16 (100) 1102 (85.4) with the length of stay. Yes 188 0 (0) 188 (14.6) Multivariable logistic regression with Preop atherosclerosis .99b stepwise selection was used to develop a No 1295 16 (100) 1279 (99.1) multivariable model to analyze the rela- Yes 12 0 (0) 12 (0.93) tionship between postoperative respirato- Preop CHF .99b ry complications and the volume of crys- No 1293 16 (100) 1277 (98.9) talloids, colloids, and total fl uids infused Yes 14 0 (0) 14 (1.08) intraoperatively. Patient age, sex, and race Preop AF .99b were the potential covariates. For all the No 1287 16 (100) 1271 (98.5) other fl uid measurements, a univariate lo- Yes 20 0 (0) 20 (1.6) b gistic regression model was used to assess Preop valve disease .99 No 1283 16 (100) 1267 (98.1) the association between each of those fl u- Yes 24 0 (0) 24 (1.9) id measures and postoperative respiratory Preop asthma .99b complication and length of stay. All tests No 1274 16 (100) 1258 (97.5) were 2-sided at a signifi cant level of .05. Yes 32 0 (0) 32 (2.5) All analyses were performed using SAS Preop COPD .99b 9.2 software (SAS Institute, Cary, North No 1299 16 (100) 1283 (99.4) Carolina).