Morbidity, Mortality, Cost, and Survival Estimates of Gastrointestinal Anastomotic Leaks Florence E Turrentine, PhD, RN, Chaderick E Denlinger, MD, Virginia B Simpson, RNNP, Robert A Garwood, MD, Stephanie Guerlain, PhD, Abhinav Agrawal, MS, Charles M Friel, Damien J LaPar, MD, MSc, George J Stukenborg, PhD, R Scott Jones, MD, FACS
MD, FACS,
Anastomotic leak, a potentially deadly postoperative occurrence, particularly interests surgeons performing gastrointestinal procedures. We investigated incidence, cost, and impact on survival of anastomotic leak in gastrointestinal surgical procedures at an academic center. STUDY DESIGN: We conducted a chart review of American College of Surgeons NSQIP operative procedures with gastrointestinal anastomosis from January 1, 2003 through April 30, 2006. Each case with an American College of Surgeons NSQIP 30-day postoperative complication was systematically reviewed for evidence of anastomotic leak for 12 months after the operative date. We tracked patients for up to 10 years to determine survival. Morbidity, mortality, and cost for patients with gastrointestinal anastomotic leaks were compared with patients with anastomoses that remained intact. RESULTS: Unadjusted analyses revealed significant differences between patients who had anastomotic leaks develop and those who did not: morbidity (98.0% vs 28.4%; p < 0.0001), length of stay (13 vs 5 days; p 0.0001), 30-day mortality (8.4% vs 2.5%; p < 0.0001), long-term mortality (36.4% vs 20.0%; p 0.0001), and hospital costs (chi-square [2] ¼ 359.8; p < 0.0001). Multivariable regression demonstrated that anastomotic leak was associated with congestive heart failure (odds ratio [OR] ¼ 31.5; 95% CI, 2.6e381.4; p ¼ 0.007), peripheral vascular disease (OR ¼ 4.6; 95% CI, 1.0e20.5; p ¼ 0.048), alcohol abuse (OR ¼ 3.7; 95% CI, 1.6e8.3; p ¼ 0.002), steroid use (OR ¼ 2.3; 95% CI: 1.1e5.0; p ¼ 0.027), abnormal sodium (OR ¼ 0.4; 95% CI, 0.2e0.7; p ¼ 0.002), weight loss (OR ¼ 0.2; 95% CI, 0.06e0.7; p ¼ 0.011), and location of anastomosis: rectum (OR ¼ 14.0; 95% CI, 2.6e75.5; p ¼ 0.002), esophagus (OR ¼ 13.0; 95% CI, 3.6e46.2; p < 0.0001), pancreas (OR ¼ 12.4; 95% CI, 3.3e46.2; p < 0.0001), small intestine (OR ¼ 6.9; 95% CI, 1.8e26.4; p ¼ 0.005), and colon (OR ¼ 5.2; 95% CI, 1.5e17.7; p ¼ 0.009). CONCLUSIONS: Significant morbidity, mortality, and cost accompany gastrointestinal anastomotic leaks. Patients who experience an anastomotic leak have lower rates of survival at 30 days and long term. (J Am Coll Surg 2015;220:195e206. 2015 by the American College of Surgeons)
BACKGROUND:
morbidity define a quality standard for surgical care. Among postoperative complications, anastomotic leaks inflict more pain and suffering than any other surgical
Surgeons pay particular attention to patient outcomes. Surgical mortality, morbidity, and effectiveness remain important professional foci. Today, risk-adjusted mortality and
Received August 19, 2014; Revised October 16, 2014; Accepted November 4, 2014. From the Department of Surgery (Turrentine, Denlinger, Simpson, Garwood, Friel, LaPar, Jones), Systems Engineering (Guerlain, Agrawal), and Public Health Sciences (Stukenborg), University of Virginia, Charlottesville, VA. Correspondence address: R Scott Jones, MD, FACS, Department of Surgery, University of Virginia, Box 800709, Charlottesville, VA 22908. email: [email protected]
CME questions for this article available at http://jacscme.facs.org Disclosure Information: Authors have nothing to disclose. Timothy J Eberlein, Editor-in-Chief, has nothing to disclose. Support: National Surgical Quality Improvement Program data was collected with funding by the Agency for Healthcare Research and Quality, Grant Number 5U18HS011913, entitled Reporting System to Improve Patient Safety in Surgery.
ª 2015 by the American College of Surgeons Published by Elsevier Inc.
195
http://dx.doi.org/10.1016/j.jamcollsurg.2014.11.002 ISSN 1072-7515/14
196
Turrentine et al
Gastrointestinal Anastomotic Leaks
complication.1 Despite continual improvement in surgical techniques, anastomotic leaks associated with gastrointestinal surgical procedures remain a major cause of postoperative mortality and morbidity.2-5 Leak rates vary depending on the site of the anastomosis: esophagus (2% to 16%),5-8 stomach (1% to 9%),9-13 pancreas (9% to 16%),14-16 bile ducts (10% to 16%),17,18 small intestine (1% to 3%),19,20 colon (3% to 29%),2,21-24 and rectum (8% to 41%),25 and have associated mortality rates as high as 35%.26 Research has identified preoperative, intraoperative, and postoperative risk factors for anastomotic leaks after specific gastrointestinal procedures.3,16,21,27 Two studies have examined long-term mortality at five28 and ten26 years post procedure. Three studies report a comprehensive review of their institution’s overall surgical leak rate,4,19,29 but we found no studies that evaluated the impact of anastomotic leak on long-term survival from all gastrointestinal procedures from mouth to anus at a single institution. We used our American College of Surgeons (ACS) NSQIP database to determine the rate of leaks at our institution and to evaluate risk factors for anastomotic leaks. We also documented the 30-day and long-term survival rates of patients who sustained an anastomotic leak. Analyzing and reporting anastomotic leak rates are important because that information can improve the quality of surgical care in the institution and can be shared with others to benchmark best practices. The aims of our study were to examine the incidence of anastomotic leaks at our institution, to identify risk factors predictive of leaks, and to explore the impact of anastomotic leaks on hospital cost and patient survival.
METHODS Our IRB approved this study. We conducted a retrospective review of prospectively collected ACS NSQIP cases entered into our institution’s database from January 1, 2003 to April 30, 2006. The details of ACS NSQIP data collection are well described in the literature.30,31 Trained surgical clinical reviewers collect the following categories of patient information for general surgery and vascular surgical procedures: demographics, selected pulmonary, cardiac, hepatobiliary, renal, vascular, central nervous system, and immunologic preoperative risk factors; operative variables and postoperative complications including wound, respiratory, renal, central nervous system, cardiovascular, and “other.” In addition, our surgical clinical reviewers (FET and VBS) collected any postoperative occurrence that occurred within 30 days of surgery. For example, atrial fibrillation, seroma formation, and ileus are not defined variables in ACS NSQIP, but were recorded by ICD-9 code32 as “other” postoperative occurrences for this study to detect any abnormalities that occurred postoperatively.
J Am Coll Surg
Sample Adult patients (18 years of age or older) undergoing major gastrointestinal operations that included an anastomosis met study criteria for inclusion. Of note, our institution participated in beta site testing for NSQIP, an Agency for Healthcare Research and Qualityefunded study, from February 2002 to September 2004. Initially, we followed the standard program collection recommendations that required the first 40 eligible cases in an 8-day cycle to be recorded. In January 1, 2003 we expanded data collection to include all eligible general surgical cases performed at the institution. Therefore, this dataset includes all ACS NSQIPeeligible gastrointestinal operations performed at our institution and not the typical ACS NSQIP restriction of 40 cases per 8-day cycle sampling. Gastrointestinal procedures, defined as CPT codes 39502 through 49999, in the Surgery/Digestive System section of the American Medical Association 2003 CPT Standard Edition33 were reviewed by a senior surgeon (RSJ) to determine if the surgical procedure included an anastomosis and would qualify for inclusion in the study. We only included esophageal cases with an abdominal incision. Each gastrointestinal procedure that included an anastomosis and had an associated 30-day postoperative complication was reviewed by a surgical resident (CED or RAG) or advanced practice surgical nurse (FET or VBS) to determine if an anastomotic leak was present. In addition, we reviewed all cases that underwent reoperation and those with an anastomosis and proximal diversion (CPT codes 44146, 44153, 44158, 44208, 44211, 44626, 45111, 45112, 45113, 45119, 45126, 45397) whether a complication was present or not, as we suspected subclinical leaks could be present in this population. Anastomotic leak was defined as a “breakdown of a suture line in a surgical anastomosis with leakage of gastric or intestinal fluid, following surgical intervention involving anastomosis of gastrointestinal or bowel structures.”34 The separation of the gastrointestinal lumen at the surgical site was identified as follows: intraoperatively by the attending surgeon during reoperation, or by the radiologist report suggesting extravasation of contrast medium at the anastomotic site, or with a clinical description of fistula at the anastomotic site, or by elevated amylase (more than twice the value for serum amylase) from pancreatic drainage. To determine if a leak was present, the patient’s electronic medical record was accessed and the operative report was reviewed to confirm an anastomosis was performed during the surgery. Next, all radiology reports, including CT, ultrasound, and fluoroscopy, were reviewed for radiologic evidence of a leak. To ensure detection of any evidence of an anastomotic leak, reports of reoperation, readmission, and
Vol. 220, No. 2, February 2015
Turrentine et al
all discharge summaries were viewed from the index operative case for 12 months after surgery, until the patient’s death, or an anastomotic leak was identified. A senior general surgeon (RSJ) or (CMF) reviewed and verified the evidence in all cases identified as containing an anastomotic leak. Consensus between clinicians reviewing the chart and the senior surgeon was reached on the presence or absence of a leak for all cases. The CPT codes related to operations performed on 7 anastomotic sites within the digestive tract (ie, esophagus, stomach, pancreas, bile ducts, small intestine, colon, and rectum) were grouped and anastomotic leak rates were then determined for each surgical site. Although not definitive, to account for the possibility that a clinically significant leak could be present without a patient exhibiting any postoperative complications collected by NSQIP, a random sample of 13.1% (n ¼ 200) of anastomotic cases without a complication were also reviewed; no leaks were identified. Variables with low rates of occurrence for individual values were combined into larger groups. Recoded preoperative variables included: American Society of Anesthesiologists physical status classification, transfusions, diabetes, race, sepsis, and functional status. The postoperative variables of wound disruption, superficial, deep, and organ space infections were combined into postoperative wound occurrences. Pneumonia, unplanned intubation, on the ventilator longer than 48 hours, and pulmonary embolus were grouped into respiratory postoperative occurrences. Urinary tract infection, acute renal failure, and progressive renal insufficiency were grouped into renal postoperative occurrences. Cerebral vascular accident, coma for longer than 24 hours, and peripheral nerve injury were grouped as central nervous system occurrences. Myocardial infarction and cardiac arrest were grouped into cardiovascular occurrences. Transfusion of >4 U packed blood cells within 72 hours after surgery, graft failure, deep venous thrombosis, systematic inflammatory response, sepsis, and septic shock were grouped into “other” occurrences. We also wanted to account for the influence that age and cancer would have on survival, so we recoded cancer to include those who received preoperative chemotherapy or radiation therapy, had disseminated cancer, or had a discharge ICD9 code reflecting malignancy. To assess surgeons’ experience we obtained the date of each surgeon’s American Board of Surgery certification. Experience was defined as time from certification to the operative date. Cost data Costs for the hospitalization of each patient in the dataset were reviewed. We compared the difference in mean costs
Gastrointestinal Anastomotic Leaks
197
for 3 groups: patients who had an intact anastomosis and did not experience a complication, those whose anastomosis was intact but did have a complication, and those who experienced an anastomotic leak. Survival We monitored patients for up to 10 years after surgery through review of our electronic medical records, institutional data repository, and Social Security Death Index to determine date of death or the last date the patient was known to be alive. Analysis The statistical significance of the bivariable relationship between the occurrence of anastomotic leak and individual patient variables was assessed using the chi-square tests for categorical variables and the 2-sample t-test and Wilcoxon rank sum test for continuous variables. Variables that demonstrated statistically significant bivariable associations (p < 0.10) were included in a multivariable logistic regression model as independent variables, with leak as the dependent variable. Kaplan-Meier analysis was used to assess the difference in survival between patients with and without leaks during the available period of follow-up. The log-rank test was used to assess the statistical significance of survival functions for each patient group. Multivariable Cox proportional hazards regression was used to measure the relative hazard of death associated with leak, adjusted for the concurrent effects of cancer and age on survival. The statistical significance of differences in mean cost associated with leaks were evaluated using the Kruskal-Wallis H test. Data analysis was conducted using IBM SPSS Statistics version 21 for Windows (IBM Corp., 2012).
RESULTS During the period from January 1, 2003 through April 30, 2006, there were 10,185 major, adult surgical procedures performed at the University of Virginia with patient data collected through ACS NSQIP. Primary and secondary CPT codes for each operative case were reviewed and 7,030 cases involved the gastrointestinal tract; 2,237 of these procedures included an anastomosis. Among these were 712 (31.8%) patients with anastomoses who had a complication as defined by ACS NSQIP within the 30day postoperative period. Demographics Table 1 lists either the median or proportion of patients with baseline characteristics for those who had an anastomotic leak develop and those who did not. Of patients undergoing gastrointestinal surgery with an anastomosis,
Turrentine et al
198
Table 1.
J Am Coll Surg
Gastrointestinal Anastomotic Leaks
Baseline Characteristics of Patients Who underwent Gastrointestinal Procedures with an Anastomosis
Risk factors
Sex, male, % Age, y, median BMI, kg/m2, median Race, white, % Emergency status, % ASA 4 or 5, % Transfer from acute care, % Wound classification, contaminated or dirty, % Diabetes, % Smoker, % Packs of cigarettes, median Alcohol consumption >2 drinks/d, % Dyspnea at rest, % DNR order, % Functional status: totally dependent, % Ventilator, % COPD, % Pneumonia, % Ascites, % Congestive heart failure, % Renal failure, % Dialysis, % Impaired sensorium, % Coma, % Hemipelgia, % Transient ischemic attack, % CVA with deficit, % CVA without deficit, % Tumor involving central nervous system, % Disseminated cancer, % Open wound, % Steroid, % Weight loss, % Bleeding disorder, % Transfusion >4 U, % Chemotherapy, % Radiotherapy, % Sepsis, % Esophageal varices, % Myocardial infarction, % Percutaneous transluminal coronary angioplasty, % Previous cardiac surgery, % Angina, % Hypertension, % Peripheral vascular disease, % Gangrene, % Abnormal sodium, %
Anastomotic leak
Anastomosis intact
51.4 59 27.2 90.7 10.3 9.3 9.3 18.7 17.8 32.7 18.9 12.2 3.7 0.93 4.7 1.87 8.4 0.93 0.93 1.87 0.93 1.87 2.8 0 0.93 0.0 3.74 2.80 0.0 0.93 3.74 12.15 9.35 3.74 0.0 0.93 5.61 4.67 0.93 2.8 7.48 7.48 0.93 53.27 2.8 0.0 32.26
39.3 51 43.6 86.7 8.0 4.5 3.9 11.8 20.02 23.1 13.3 2.3 1.8 0.75 2.4 0.99 3.8 0.14 0.47 0.33 0.14 0.75 0.8 0 0.38 1.88 1.97 2.21 0.0 3.81 2.21 6.86 6.81 0.8 0.33 1.32 3.38 2.77 0.14 0.47 3.24 3.43 0.38 46.85 0.94 0.09 14.67
p Value
0.0121* 0.01* 0.0001* 0.30 0.41 0.0064* 0.006* 0.033* 0.57 0.0226* 0.152
MD, FACS,
Anastomotic leak, a potentially deadly postoperative occurrence, particularly interests surgeons performing gastrointestinal procedures. We investigated incidence, cost, and impact on survival of anastomotic leak in gastrointestinal surgical procedures at an academic center. STUDY DESIGN: We conducted a chart review of American College of Surgeons NSQIP operative procedures with gastrointestinal anastomosis from January 1, 2003 through April 30, 2006. Each case with an American College of Surgeons NSQIP 30-day postoperative complication was systematically reviewed for evidence of anastomotic leak for 12 months after the operative date. We tracked patients for up to 10 years to determine survival. Morbidity, mortality, and cost for patients with gastrointestinal anastomotic leaks were compared with patients with anastomoses that remained intact. RESULTS: Unadjusted analyses revealed significant differences between patients who had anastomotic leaks develop and those who did not: morbidity (98.0% vs 28.4%; p < 0.0001), length of stay (13 vs 5 days; p 0.0001), 30-day mortality (8.4% vs 2.5%; p < 0.0001), long-term mortality (36.4% vs 20.0%; p 0.0001), and hospital costs (chi-square [2] ¼ 359.8; p < 0.0001). Multivariable regression demonstrated that anastomotic leak was associated with congestive heart failure (odds ratio [OR] ¼ 31.5; 95% CI, 2.6e381.4; p ¼ 0.007), peripheral vascular disease (OR ¼ 4.6; 95% CI, 1.0e20.5; p ¼ 0.048), alcohol abuse (OR ¼ 3.7; 95% CI, 1.6e8.3; p ¼ 0.002), steroid use (OR ¼ 2.3; 95% CI: 1.1e5.0; p ¼ 0.027), abnormal sodium (OR ¼ 0.4; 95% CI, 0.2e0.7; p ¼ 0.002), weight loss (OR ¼ 0.2; 95% CI, 0.06e0.7; p ¼ 0.011), and location of anastomosis: rectum (OR ¼ 14.0; 95% CI, 2.6e75.5; p ¼ 0.002), esophagus (OR ¼ 13.0; 95% CI, 3.6e46.2; p < 0.0001), pancreas (OR ¼ 12.4; 95% CI, 3.3e46.2; p < 0.0001), small intestine (OR ¼ 6.9; 95% CI, 1.8e26.4; p ¼ 0.005), and colon (OR ¼ 5.2; 95% CI, 1.5e17.7; p ¼ 0.009). CONCLUSIONS: Significant morbidity, mortality, and cost accompany gastrointestinal anastomotic leaks. Patients who experience an anastomotic leak have lower rates of survival at 30 days and long term. (J Am Coll Surg 2015;220:195e206. 2015 by the American College of Surgeons)
BACKGROUND:
morbidity define a quality standard for surgical care. Among postoperative complications, anastomotic leaks inflict more pain and suffering than any other surgical
Surgeons pay particular attention to patient outcomes. Surgical mortality, morbidity, and effectiveness remain important professional foci. Today, risk-adjusted mortality and
Received August 19, 2014; Revised October 16, 2014; Accepted November 4, 2014. From the Department of Surgery (Turrentine, Denlinger, Simpson, Garwood, Friel, LaPar, Jones), Systems Engineering (Guerlain, Agrawal), and Public Health Sciences (Stukenborg), University of Virginia, Charlottesville, VA. Correspondence address: R Scott Jones, MD, FACS, Department of Surgery, University of Virginia, Box 800709, Charlottesville, VA 22908. email: [email protected]
CME questions for this article available at http://jacscme.facs.org Disclosure Information: Authors have nothing to disclose. Timothy J Eberlein, Editor-in-Chief, has nothing to disclose. Support: National Surgical Quality Improvement Program data was collected with funding by the Agency for Healthcare Research and Quality, Grant Number 5U18HS011913, entitled Reporting System to Improve Patient Safety in Surgery.
ª 2015 by the American College of Surgeons Published by Elsevier Inc.
195
http://dx.doi.org/10.1016/j.jamcollsurg.2014.11.002 ISSN 1072-7515/14
196
Turrentine et al
Gastrointestinal Anastomotic Leaks
complication.1 Despite continual improvement in surgical techniques, anastomotic leaks associated with gastrointestinal surgical procedures remain a major cause of postoperative mortality and morbidity.2-5 Leak rates vary depending on the site of the anastomosis: esophagus (2% to 16%),5-8 stomach (1% to 9%),9-13 pancreas (9% to 16%),14-16 bile ducts (10% to 16%),17,18 small intestine (1% to 3%),19,20 colon (3% to 29%),2,21-24 and rectum (8% to 41%),25 and have associated mortality rates as high as 35%.26 Research has identified preoperative, intraoperative, and postoperative risk factors for anastomotic leaks after specific gastrointestinal procedures.3,16,21,27 Two studies have examined long-term mortality at five28 and ten26 years post procedure. Three studies report a comprehensive review of their institution’s overall surgical leak rate,4,19,29 but we found no studies that evaluated the impact of anastomotic leak on long-term survival from all gastrointestinal procedures from mouth to anus at a single institution. We used our American College of Surgeons (ACS) NSQIP database to determine the rate of leaks at our institution and to evaluate risk factors for anastomotic leaks. We also documented the 30-day and long-term survival rates of patients who sustained an anastomotic leak. Analyzing and reporting anastomotic leak rates are important because that information can improve the quality of surgical care in the institution and can be shared with others to benchmark best practices. The aims of our study were to examine the incidence of anastomotic leaks at our institution, to identify risk factors predictive of leaks, and to explore the impact of anastomotic leaks on hospital cost and patient survival.
METHODS Our IRB approved this study. We conducted a retrospective review of prospectively collected ACS NSQIP cases entered into our institution’s database from January 1, 2003 to April 30, 2006. The details of ACS NSQIP data collection are well described in the literature.30,31 Trained surgical clinical reviewers collect the following categories of patient information for general surgery and vascular surgical procedures: demographics, selected pulmonary, cardiac, hepatobiliary, renal, vascular, central nervous system, and immunologic preoperative risk factors; operative variables and postoperative complications including wound, respiratory, renal, central nervous system, cardiovascular, and “other.” In addition, our surgical clinical reviewers (FET and VBS) collected any postoperative occurrence that occurred within 30 days of surgery. For example, atrial fibrillation, seroma formation, and ileus are not defined variables in ACS NSQIP, but were recorded by ICD-9 code32 as “other” postoperative occurrences for this study to detect any abnormalities that occurred postoperatively.
J Am Coll Surg
Sample Adult patients (18 years of age or older) undergoing major gastrointestinal operations that included an anastomosis met study criteria for inclusion. Of note, our institution participated in beta site testing for NSQIP, an Agency for Healthcare Research and Qualityefunded study, from February 2002 to September 2004. Initially, we followed the standard program collection recommendations that required the first 40 eligible cases in an 8-day cycle to be recorded. In January 1, 2003 we expanded data collection to include all eligible general surgical cases performed at the institution. Therefore, this dataset includes all ACS NSQIPeeligible gastrointestinal operations performed at our institution and not the typical ACS NSQIP restriction of 40 cases per 8-day cycle sampling. Gastrointestinal procedures, defined as CPT codes 39502 through 49999, in the Surgery/Digestive System section of the American Medical Association 2003 CPT Standard Edition33 were reviewed by a senior surgeon (RSJ) to determine if the surgical procedure included an anastomosis and would qualify for inclusion in the study. We only included esophageal cases with an abdominal incision. Each gastrointestinal procedure that included an anastomosis and had an associated 30-day postoperative complication was reviewed by a surgical resident (CED or RAG) or advanced practice surgical nurse (FET or VBS) to determine if an anastomotic leak was present. In addition, we reviewed all cases that underwent reoperation and those with an anastomosis and proximal diversion (CPT codes 44146, 44153, 44158, 44208, 44211, 44626, 45111, 45112, 45113, 45119, 45126, 45397) whether a complication was present or not, as we suspected subclinical leaks could be present in this population. Anastomotic leak was defined as a “breakdown of a suture line in a surgical anastomosis with leakage of gastric or intestinal fluid, following surgical intervention involving anastomosis of gastrointestinal or bowel structures.”34 The separation of the gastrointestinal lumen at the surgical site was identified as follows: intraoperatively by the attending surgeon during reoperation, or by the radiologist report suggesting extravasation of contrast medium at the anastomotic site, or with a clinical description of fistula at the anastomotic site, or by elevated amylase (more than twice the value for serum amylase) from pancreatic drainage. To determine if a leak was present, the patient’s electronic medical record was accessed and the operative report was reviewed to confirm an anastomosis was performed during the surgery. Next, all radiology reports, including CT, ultrasound, and fluoroscopy, were reviewed for radiologic evidence of a leak. To ensure detection of any evidence of an anastomotic leak, reports of reoperation, readmission, and
Vol. 220, No. 2, February 2015
Turrentine et al
all discharge summaries were viewed from the index operative case for 12 months after surgery, until the patient’s death, or an anastomotic leak was identified. A senior general surgeon (RSJ) or (CMF) reviewed and verified the evidence in all cases identified as containing an anastomotic leak. Consensus between clinicians reviewing the chart and the senior surgeon was reached on the presence or absence of a leak for all cases. The CPT codes related to operations performed on 7 anastomotic sites within the digestive tract (ie, esophagus, stomach, pancreas, bile ducts, small intestine, colon, and rectum) were grouped and anastomotic leak rates were then determined for each surgical site. Although not definitive, to account for the possibility that a clinically significant leak could be present without a patient exhibiting any postoperative complications collected by NSQIP, a random sample of 13.1% (n ¼ 200) of anastomotic cases without a complication were also reviewed; no leaks were identified. Variables with low rates of occurrence for individual values were combined into larger groups. Recoded preoperative variables included: American Society of Anesthesiologists physical status classification, transfusions, diabetes, race, sepsis, and functional status. The postoperative variables of wound disruption, superficial, deep, and organ space infections were combined into postoperative wound occurrences. Pneumonia, unplanned intubation, on the ventilator longer than 48 hours, and pulmonary embolus were grouped into respiratory postoperative occurrences. Urinary tract infection, acute renal failure, and progressive renal insufficiency were grouped into renal postoperative occurrences. Cerebral vascular accident, coma for longer than 24 hours, and peripheral nerve injury were grouped as central nervous system occurrences. Myocardial infarction and cardiac arrest were grouped into cardiovascular occurrences. Transfusion of >4 U packed blood cells within 72 hours after surgery, graft failure, deep venous thrombosis, systematic inflammatory response, sepsis, and septic shock were grouped into “other” occurrences. We also wanted to account for the influence that age and cancer would have on survival, so we recoded cancer to include those who received preoperative chemotherapy or radiation therapy, had disseminated cancer, or had a discharge ICD9 code reflecting malignancy. To assess surgeons’ experience we obtained the date of each surgeon’s American Board of Surgery certification. Experience was defined as time from certification to the operative date. Cost data Costs for the hospitalization of each patient in the dataset were reviewed. We compared the difference in mean costs
Gastrointestinal Anastomotic Leaks
197
for 3 groups: patients who had an intact anastomosis and did not experience a complication, those whose anastomosis was intact but did have a complication, and those who experienced an anastomotic leak. Survival We monitored patients for up to 10 years after surgery through review of our electronic medical records, institutional data repository, and Social Security Death Index to determine date of death or the last date the patient was known to be alive. Analysis The statistical significance of the bivariable relationship between the occurrence of anastomotic leak and individual patient variables was assessed using the chi-square tests for categorical variables and the 2-sample t-test and Wilcoxon rank sum test for continuous variables. Variables that demonstrated statistically significant bivariable associations (p < 0.10) were included in a multivariable logistic regression model as independent variables, with leak as the dependent variable. Kaplan-Meier analysis was used to assess the difference in survival between patients with and without leaks during the available period of follow-up. The log-rank test was used to assess the statistical significance of survival functions for each patient group. Multivariable Cox proportional hazards regression was used to measure the relative hazard of death associated with leak, adjusted for the concurrent effects of cancer and age on survival. The statistical significance of differences in mean cost associated with leaks were evaluated using the Kruskal-Wallis H test. Data analysis was conducted using IBM SPSS Statistics version 21 for Windows (IBM Corp., 2012).
RESULTS During the period from January 1, 2003 through April 30, 2006, there were 10,185 major, adult surgical procedures performed at the University of Virginia with patient data collected through ACS NSQIP. Primary and secondary CPT codes for each operative case were reviewed and 7,030 cases involved the gastrointestinal tract; 2,237 of these procedures included an anastomosis. Among these were 712 (31.8%) patients with anastomoses who had a complication as defined by ACS NSQIP within the 30day postoperative period. Demographics Table 1 lists either the median or proportion of patients with baseline characteristics for those who had an anastomotic leak develop and those who did not. Of patients undergoing gastrointestinal surgery with an anastomosis,
Turrentine et al
198
Table 1.
J Am Coll Surg
Gastrointestinal Anastomotic Leaks
Baseline Characteristics of Patients Who underwent Gastrointestinal Procedures with an Anastomosis
Risk factors
Sex, male, % Age, y, median BMI, kg/m2, median Race, white, % Emergency status, % ASA 4 or 5, % Transfer from acute care, % Wound classification, contaminated or dirty, % Diabetes, % Smoker, % Packs of cigarettes, median Alcohol consumption >2 drinks/d, % Dyspnea at rest, % DNR order, % Functional status: totally dependent, % Ventilator, % COPD, % Pneumonia, % Ascites, % Congestive heart failure, % Renal failure, % Dialysis, % Impaired sensorium, % Coma, % Hemipelgia, % Transient ischemic attack, % CVA with deficit, % CVA without deficit, % Tumor involving central nervous system, % Disseminated cancer, % Open wound, % Steroid, % Weight loss, % Bleeding disorder, % Transfusion >4 U, % Chemotherapy, % Radiotherapy, % Sepsis, % Esophageal varices, % Myocardial infarction, % Percutaneous transluminal coronary angioplasty, % Previous cardiac surgery, % Angina, % Hypertension, % Peripheral vascular disease, % Gangrene, % Abnormal sodium, %
Anastomotic leak
Anastomosis intact
51.4 59 27.2 90.7 10.3 9.3 9.3 18.7 17.8 32.7 18.9 12.2 3.7 0.93 4.7 1.87 8.4 0.93 0.93 1.87 0.93 1.87 2.8 0 0.93 0.0 3.74 2.80 0.0 0.93 3.74 12.15 9.35 3.74 0.0 0.93 5.61 4.67 0.93 2.8 7.48 7.48 0.93 53.27 2.8 0.0 32.26
39.3 51 43.6 86.7 8.0 4.5 3.9 11.8 20.02 23.1 13.3 2.3 1.8 0.75 2.4 0.99 3.8 0.14 0.47 0.33 0.14 0.75 0.8 0 0.38 1.88 1.97 2.21 0.0 3.81 2.21 6.86 6.81 0.8 0.33 1.32 3.38 2.77 0.14 0.47 3.24 3.43 0.38 46.85 0.94 0.09 14.67
p Value
0.0121* 0.01* 0.0001* 0.30 0.41 0.0064* 0.006* 0.033* 0.57 0.0226* 0.152
