Surgical site infections (SSIs) occur in 2%–5% of patients having inpatient surgery and are associated with a 2- to 11-fold increase in mortality.1 SSIs increase the length of hospitalization after surgery by nearly a week and lead to an additional $10 billion in health care spending per year in the United States.1 Patients having colon resection surgery are at particularly high risk for SSIs because of the high degree of bacterial colonization in the colon and the potential for anastomotic leakage after surgery. In one large prospective cohort of colon surgery patients, SSIs occurred in 4% of patients.2
Red blood cell (RBC) transfusion is given to patients during the perioperative period to increase tissue oxygen delivery. Approximately 13 million patients per year receive RBC transfusion in the United States, and blood transfusion practices continue to vary highly by center.3–5 In a recent large epidemiologic study of more than 300,000 hospitalized patients, 11% received RBC transfusion.6 Although RBC transfusion can restore blood hemoglobin levels toward normal values and can increase oxygen-carrying capacity of blood, it may also increase the risk of infection. In one study of more than 2000 critically ill patients, RBC transfusion increased the risk of nosocomial infection by 8%.7 In a second study of 125,000 general surgery patients, transfusion of 1–2 RBC units during surgery increased the odds of postoperative pneumonia, death, and sepsis significantly, even after adjusting for confounders using propensity score analysis.8
The association between RBC transfusion and SSI after colon surgery has not been investigated rigorously. In a cohort study of 150 patients undergoing abdominoperineal resection of the colon, RBC transfusion was associated with a 3-fold increase in SSI risk.9 This study used multivariable regression to control for potential confounders, but did not analyze covariate balance in the transfused and nontransfused groups after adjustment. The purpose of our study was to investigate RBC transfusion as an independent risk factor for SSI in a large cohort of colon resection patients. We hypothesized that RBC transfusion would be associated with increased risk for SSI.
The study was exempt from institutional review board approval because it was not human subject research.
The strobe checklist for cohort studies was referenced when preparing the manuscript. Study design, outcome variables, and analysis plan were selected before performing the data analysis.
The 2014 National Surgical Quality Improvement Program (NSQIP) participant use file (PUF) was queried to identify patients who had colon resection surgery. The NSQIP PUF contains data on surgical patients from all participating hospitals (more than 700 centers) and is reported for each year of the program. Patients were identified using current procedural terminology (CPT) codes. The following CPT codes were used to identify patients for inclusion in the study: 44140, 44141, 44143, 44144, 44146, 44147, 44150, 44151, 44155, 44156, 44157, 44158, 44160, 44204, 44205, 44206, 44207, 44208, 44210, 44211, and 44212. Patients were excluded from the data set if they were: >75 years of age, had an open/infected surgical site before surgery, or were having emergency surgery. Patients were also excluded if they were missing data for any study variable.
Study variables were defined according to the 2014 NSQIP PUF user guide (www.facs.org). The following study variables were collected from the PUF because they were considered to be potential risk factors for SSI: age, sex, body weight, American Society of Anesthesiologists (ASA) functional class, diabetes mellitus, tobacco use, chronic obstructive pulmonary disease, congestive heart failure, hypertension, acute renal failure, dialysis, disseminated cancer, steroid use, preoperative sepsis, ascites, preoperative hematocrit, open wound, wound class, and total operative time.
The primary exposure of interest was perioperative RBC transfusion, which is defined in the NSQIP PUF as transfusion of at least 1 unit of RBCs or an equivalent amount of autologous blood within 72 hours of surgery start. Data were also collected on the timing of perioperative RBC transfusion.
All SSIs were defined according to NSQIP definitions, which are available on the American College of Surgeons website in the NSQIP PUF user guide (www.facs.org). The following SSIs were analyzed as primary outcome variables and were selected before data analysis: superficial incisional SSI, deep incisional SSI, organ space SSI, and postoperative septic shock. NSQIP defines superficial incisional SSI as infection that occurs within 30 days of surgery and involves only the skin or subcutaneous tissues and has one confirmatory finding (eg, purulent drainage, pathogenic bacterial growth, or wound erythema requiring intentional wound reopening). Deep surgical SSI is defined as infection that occurs within 30 days of surgery in the muscle or fascia and has one confirmatory finding (purulent drainage, abscess, or wound dehiscence with fever or localized pain). Organ space infection is defined as infection that occurs within 30 days of surgery and involves part of the surgical anatomy other than the incision with one confirmatory finding (purulence, pathogenic bacterial growth, or abscess). Septic shock is defined as infection with associated circulatory dysfunction (hypotension or vasopressor use). Data were also collected on the timing of SSIs and septic shock.
The study sample size was based upon the number of cases that were available in the 2014 NSQIP PUF after accounting for study inclusion and exclusion criteria. Given that approximately 10% of patients in the cohort were transfused and assuming a baseline risk for SSI of 3%,10 the available sample size allowed for determination of a 2-fold increase in the odds of SSI with 90% power and an α of .01.
Statistical analysis was performed using SAS 9.3 (SAS Corporation, Cary, NC). Baseline characteristics were summarized as the mean value ± standard deviation or number and percentage of patients. Baseline variables were compared between patients who were transfused and those who were not using either Student t test for continuous variables or the χ2 test for categorical variables. Standardized differences were also calculated. The timing of RBC transfusion and SSIs was explored using box and whisker plots.
To control for differences in baseline characteristics, which could affect SSI risk, a propensity score model for RBC transfusion was created. Independent variables were selected for inclusion in the model according to criteria suggested by Austin: variables were associated with both the exposure and outcome or only the outcome. The propensity score model was specified so that RBC transfusion was the dependent variable. Independent variables in the model included: age, sex, body weight, ASA functional status, diabetes mellitus, tobacco use, chronic obstructive pulmonary disease, ascites, congestive heart failure, hypertension, acute renal failure, dialysis, disseminated cancer, steroid use, preoperative sepsis, open wound, preoperative hematocrit, operative time, and wound class. The ultimate goal of the propensity score model was to balance baseline covariates that could affect SSI risk and differ by exposure group.11 After the propensity score model was specified, inverse probability of treatment weights (IPTW) were calculated for the exposed group as 1/propensity score and for the control group as 1/(1−propensity score). Normalized IPTWs were calculated by dividing the individual IPTW by the mean IPTW.
The propensity score model was assessed by comparing baseline covariates between exposure groups after weighting with normalized IPTWs. Standardized differences were calculated for covariates, and a standardized difference <0.1 was considered to represent good covariate balance based on recommendations in the published literature.10 Because baseline covariates were not initially well balanced, truncation of the cohort was performed from the extreme ends of the IPTW distribution until baseline covariates were better balanced between the groups. Only patients with an IPTW that fell between the 25th and 75th percentile were ultimately included in the adjusted risk estimations.
To estimate the associations between SSIs and RBC transfusion, weighted logistic regression was performed using normalized IPTWs. Odds ratios (ORs) with 99% confidence intervals (CIs) were reported for superficial and deep incisional SSI, body cavity SSI, and septic shock. For all analyses, a P value less than .01 was considered statistically significant after accounting for multiple statistical tests.
Of the 23,388 patients who met study inclusion criteria, 1845 (7.9%) patients received RBC transfusion during surgery or within 72 hours after surgery. The median day of RBC transfusion was postoperative day 1 (interquartile range [IQR] = 0–2). 1284 patients (5.1%) in the cohort developed a superficial incisional SSI, 1806 patients (7.2%) developed a deep incisional SSI, 1072 patients (4.3%) developed an organ space SSI, and 344 (1.4%) developed septic shock. The median day for superficial incisional SSI was day 10 (IQR = 7–15), for deep incisional SSI, it was day 11 (IQR = 8–17), for organ space, SSI was day 10 (IQR = 6–16), and for septic shock, day 6 (IQR = 2–10).
Baseline characteristic for patients who were transfused and for those who were not transfused are shown in Table 1. The 2 groups were highly imbalanced, with transfused patients being older, having more comorbidity, more steroid use, longer operative times, and more contaminated wounds. Propensity score analysis and IPTW adjustment created balance in all measured baseline covariates except for age, body weight, and hematocrit (Table 2). Even after IPTW adjustment, patients who were transfused had a mean hematocrit that was 1 point lower than those who were not transfused.
Table 3 shows IPTW adjusted rates and ORs for SSIs in patients who received RBC transfusion. There was no apparent difference in superficial (OR, 1.18; 99% CI, 0.48–2.88) or deep (OR, 1.47; 99% CI, 0.23–9.43) incisional SSI between those who were transfused and those who were not (P = .77 and .66 respectively); however, CIs around the OR point estimates were wide. Patients who were transfused appeared to have more organ space SSIs (OR, 2.93; 99% CI, 1.43–6.01) and more septic shock (OR, 9.23; 99% CI, 3.53–24.09; both P < .0001), but the CIs around the OR point estimate were wide.
Transfusion of RBCs in surgical patients with major bleeding can be a lifesaving intervention, but RBC transfusion is also associated with numerous complications, including infection.12–14 Due to concerns about transfusion-related complications and a lack of evidence demonstrating benefit for RBC transfusion in many settings, recent guidelines have endorsed restrictive RBC transfusion thresholds in all patients except for those with major hemorrhage or acute coronary syndromes.15
RBC transfusion has the potential to increase infection risk for several reasons, and the term transfusion-related immunomodulation has recently been used to describe the immunologic changes that occur with allogeneic blood transfusion. These changes included impairment of normal monocyte function and decreased tumor necrosis α production in response to endotoxin (lipopolysaccharide).16,17 The pathophysiology of transfusion-related immunomodulation is not fully understood, but it is thought to be related to both the RBC storage lesion as well as RBC storage solutions.16
Currently, RBCs can be refrigerated for up to 42 days. During this period, numerous changes occur in RBCs, including depletion of adenine triphosphate and 2,3-diphosphoglycerate, formation of hemoglobin containing microparticles, and release of extracellular hemoglobin and potassium.18,19 Extracellular hemoglobin levels can reach as high as 80 μM, when RBCs are stored for 50 days, which is comparable to levels found in sickle cell anemia patients.18 Extracellular hemoglobin has numerous adverse effects, including reduction in bioavailable nitric oxide, which can lead to microcirculatory dysfunction and tissue malperfusion.
Animal studies also suggest that transfusion of stored RBCs increases inflammation. RBC microparticles, which are common in banked RBCs, interact with platelets to release chemokines.20 Microparticles also express phosphatidylserine on their surface, which increases thrombin generation and activates the innate immune system.20,21 Transfused RBCs can adhere to endothelial cells, which does not occur in the normal vasculature to a significant degree.22 The significance of red cell adhesion is not known, but it may decrease tissue oxygen delivery, which could affect wound healing. Finally, transfusion of RBCs increases serum iron levels substantially, which can promote bacterial growth and increase infectious risk in surgical patients.23,24
In our study, RBC transfusion had no apparent association with superficial or deep incisional SSI, but appeared to be associated with organ space infections and septic shock. The implication of this finding is that colon surgery patients who receive RBC transfusion may be at risk for anastomotic breakdown, which can lead to organ space infection and sepsis. Surgical anastomotic sites typically have marginal blood flow compared with intact tissues immediately after surgery.25 Tightly stapled anastomoses may be particularly vulnerable and subject to poor healing.25 In our cohort, patients who received RBC transfusion appeared to have a 9-fold increase in septic shock after surgery, which could have important implications for postoperative monitoring and perhaps the duration of antibiotic prophylaxis.
Preoperative anemia could also be a critical risk factor for RBC transfusion in colon surgery patients and may contribute to postoperative infection risk. In both gynecologic surgery patients and general surgery patients, preoperative anemia and intraoperative RBC transfusion have been shown to be independent risk factors for postoperative infection.26,27 At the present time, there is limited evidence to support anemia treatment before colorectal surgery, but a recent meta-analysis suggests that iron supplementation in patients with iron deficiency anemia might decrease the number of patients receiving RBC transfusion during gastrointestinal surgery.28 Also, a meta-analysis of 21 randomized trials suggests that acceptance of a lower RBC transfusion threshold decreases the risk of hospital-acquired infection by approximately 5%, and that for every 38 patients without RBC transfusion, one infection is prevented.29 Future studies in colon surgery patients should explore whether preoperative anemia treatment and restrictive transfusion practices can help to decrease RBC transfusion and SSI risk.
One important limitation of our study is that the NSQIP PUF defines perioperative transfusion as receiving either allogeneic RBCs or an equivalent amount of autologous or cell saver blood. Further, the PUF does not specify whether autologous blood was banked before transfusion or collected from the surgical field and then immediately transfused back to the patient. Unfortunately, there is no way to discern which patients received autologous blood versus allogeneic blood in the PUF. If a significant number of patients received fresh autologous blood, they should not be at risk for storage-related RBC transfusion issues, suggesting an alternative explanation for the association between RBC transfusion and SSI.
Our study also has other limitations. First, it is retrospective and cannot establish causality. Although we used propensity score analysis and IPTW adjustment to improve balance in baseline covariates, not all covariates could be fully balanced between exposure groups (eg, age, body weight, and preoperative hematocrit). It is possible that important unmeasured confounders might have been present in both exposure groups; thus, residual confounding might have biased our risk estimates. The most salient of these is that there is no easy way to control for surgical complexity or surgeon skill level in the NSQIP PUF. It is possible that surgeons who have bleeding complications are also more likely to have anastomotic complications, which we could not control for in the analysis. It seems unlikely that surgical complexity can explain our findings entirely, though, because surgical duration was not different by exposure group in the IPTW-adjusted cohort. Another limitation of our findings is that the 2014 NSQIP PUF does not contain data on the total amount of RBC transfusion or transfusion of other blood products such as platelets, which have the highest rate of bacterial contamination. Instead, perioperative RBC transfusion is recorded as a dichotomous variable representing any RBC transfusion that occurred during the 72 hours after surgical incision. It is also possible that there is information bias (nondifferential misclassification bias) in our study because of the definitions that NSQIP uses for SSI. Although some studies have debated NSQIP SSI definitions, NSQIP is a well-described and externally validated clinical database with data abstracted by trained clinicians. It has also been used as a “gold standard” when comparing SSI definitions.30 Finally, some studies have suggested that old transfused RBCs are associated with more complications than fresh RBCs, and the NSQIP PUF has no data on the age of RBCs.31 If patients in the cohort received mostly old blood, the study results may not be generalizable to all patient populations.
In summary, in a large contemporary cohort of colon resection patients, perioperative RBC transfusion had no apparent association with an increased risk for superficial or deep incisional SSI, but appeared to be associated with a 3-fold increase in the risk of organ space infections and a 9-fold increase in septic shock. In colon resection patients, preoperative anemia management may be critical, and restrictive RBC transfusion practices may be merited to mitigate these risks. Future studies are needed to confirm our findings.
The authors would like to thank the American College of Surgeons and participating NSQIP centers for the data collection that allowed us to perform this study.
The American College of Surgeons National Surgical Quality Improvement Program and the hospitals participating in the ACS NSQIP are the source of the data used herein; they have not verified and are not responsible for the statistical validity of the data analysis or the conclusions derived by the authors.
Name: Michael Mazzeffi, MD, MPH, MSc.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Name: Kenichi Tanaka, MD, MSc.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Name: Samuel Galvagno, DO, PhD, FCCM.
Contribution: This author helped design the study, analyze the data, and write the manuscript.
This manuscript was handled by: Marisa B. Marques, MD.
2. Tang R, Chen HH, Wang YL, et al. Risk factors for surgical site infection after elective resection of the colon and rectum: a single-center prospective study of 2,809 consecutive patients. Ann Surg. 2001;234:181–189.
3. Arya RC, Wander G, Gupta P. Blood component therapy: which, when and how much. J Anaesthesiol Clin Pharmacol. 2011;27:278–284.
4. Snyder-Ramos SA, Möhnle P, Weng YS, et al.; Investigators of the Multicenter Study of Perioperative Ischemia; MCSPI Research Group. The ongoing variability in blood transfusion practices in cardiac surgery. Transfusion. 2008;48:1284–1299.
6. Roubinian NH, Escobar GJ, Liu V, et al.; NHLBI Recipient Epidemiology and Donor Evaluation Study (REDS-III). Trends in red blood cell transfusion and 30-day mortality among hospitalized patients. Transfusion. 2014;54:2678–2686.
7. Taylor RW, O’Brien J, Trottier SJ, et al. Red blood cell transfusions and nosocomial infections in critically ill patients. Crit Care Med. 2006;34:2302–2308.
8. Bernard AC, Davenport DL, Chang PK, Vaughan TB, Zwischenberger JB. Intraoperative transfusion of 1 U to 2 U packed red blood cells is associated with increased 30-day mortality, surgical-site infection, pneumonia, and sepsis in general surgery patients. J Am Coll Surg. 2009;208:931–937.
9. Kaneko K, Kawai K, Tsuno NH, et al. Perioperative allogeneic blood transfusion is associated with surgical site infection after abdominoperineal resection-a space for the implementation of patient blood management strategies. Int Surg. 2015;100:797–804.
10. Maojun G, Lewis SS, Moehring RW, et al. Rates of surgical site infection after colon surgery: a comparison of outcomes using a laparoscopic approach compared to open operations. Open Forum Infect Dis. 2016;3(suppl 1):1457.
11. Austin PC, Stuart EA. Moving towards best practice when using inverse probability of treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in observational studies. Stat Med. 2015;34:3661–3679.
12. de Lissovoy G, Fraeman K, Hutchins V, Murphy D, Song D, Vaughn BB. Surgical site infection: incidence and impact on hospital utilization and treatment costs. Am J Infect Control. 2009;37:387–397.
13. Kirkland KB, Briggs JP, Trivette SL, Wilkinson WE, Sexton DJ. The impact of surgical-site infections in the 1990s: attributable mortality, excess length of hospitalization, and extra costs. Infect Control Hosp Epidemiol. 1999;20:725–730.
14. Anthony T, Long J, Hynan LS, et al. Surgical complications exert a lasting effect on disease-specific health-related quality of life for patients with colorectal cancer. Surgery. 2003;134:119–125.
15. Alexander J, Cifu AS. Transfusion of red blood cells. JAMA. 2016;316:2038–2039.
16. Muszynski J, Nateri J, Nicol K, Greathouse K, Hanson L, Hall M. Immunosuppressive effects of red blood cells on monocytes are related to both storage time and storage solution. Transfusion. 2012;52:794–802.
17. Muszynski JA, Frazier E, Nofziger R, et al.; Pediatric Critical Care Blood Research Network (Blood Net) Subgroup of the Pediatric Acute Lung Injury and Sepsis Investigators (PALISI). Red blood cell transfusion and immune function in critically ill children: a prospective observational study. Transfusion. 2015;55:766–774.
18. Kim-Shapiro DB, Lee J, Gladwin MT. Storage lesion: role of red blood cell breakdown. Transfusion. 2011;51:844–851.
19. Salzer U, Zhu R, Luten M, et al. Vesicles generated during storage of red cells are rich in the lipid raft marker stomatin. Transfusion. 2008;48:451–462.
20. Xiong Z, Cavaretta J, Qu L, Stolz DB, Triulzi D, Lee JS. Red blood cell microparticles show altered inflammatory chemokine binding and release ligand upon interaction with platelets. Transfusion. 2011;51:610–621.
21. Wang RH, Phillips G Jr, Medof ME, Mold C. Activation of the alternative complement pathway by exposure of phosphatidylethanolamine and phosphatidylserine on erythrocytes from sickle cell disease patients. J Clin Invest. 1993;92:1326–1335.
22. Luk CS, Gray-Statchuk LA, Cepinkas G, Chin-Yee IH. WBC reduction reduces storage-associated RBC adhesion to human vascular endothelial cells under conditions of continuous flow in vitro. Transfusion. 2003;43:151–156.
23. Saxena S, Shulman IA, Johnson C. Effect of blood transfusion on serum iron and transferrin saturation. Arch Pathol Lab Med. 1993;117:622–624.
24. Chow JK, Werner BG, Ruthazer R, Snydman DR. Increased serum iron levels and infectious complications after liver transplantation. Clin Infect Dis. 2010;51:e16–e23.
25. Chung RS. Blood flow in colonic anastomoses. Effect of stapling and suturing. Ann Surg. 1987;206:335–339.
26. Richards T, Musallam KM, Nassif J, et al. Impact of preoperative anaemia and blood transfusion on postoperative outcomes in gynaecological surgery. PLoS One. 2015;10:e0130861.
27. Glance LG, Dick AW, Mukamel DB, et al. Association between intraoperative blood transfusion and mortality and morbidity in patients undergoing noncardiac surgery. Anesthesiology. 2011;114:283–292.
28. Hallet J, Hanif A, Callum J, et al. The impact of perioperative iron on the use of red blood cell transfusions in gastrointestinal surgery: a systematic review and meta-analysis. Transfus Med Rev. 2014;28:205–211.
29. Rohde JM, Dimcheff DE, Blumberg N, et al. Health care-associated infection after red blood cell transfusion: a systematic review and meta-analysis. JAMA. 2014;311:1317–1326.
30. Ju MH, Ko CY, Hall BL, Bosk CL, Bilimoria KY, Wick EC. A comparison of 2 surgical site infection monitoring systems. JAMA Surg. 2015;150:51–57.
31. Koch CG, Li L, Sessler DI, et al. Duration of red-cell storage and complications after cardiac surgery. N Engl J Med. 2008;358:1229–1239.