Preoperative anemia and postoperative outcomes after hepatectomy

Transcription

1 HPB ORIGINAL ARTICLE Preoperative anemia and postoperative outcomes after hepatectomy Samer Tohme 1, Patrick R. Varley 1, Douglas P. Landsittel 2, Alexis P. Chidi 1 & Allan Tsung 1 1 Department of Surgery, and 2 Department of General Internal Medicine, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA Abstract Background: Preoperative anaemia is associated with adverse outcomes after surgery but outcomes after liver surgery specifically are not well established. We aimed to analyze the incidence of and effects of preoperative anemia on morbidity and mortality in patients undergoing liver resection. Methods: All elective hepatectomies performed for the period recorded in the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) database were evaluated. We obtained anonymized data for 30-day mortality and major morbidity (one or more major complication), demographics, and preoperative and perioperative risk factors. We used multivariable logistic regression models to assess the adjusted effect of anemia, which was defined as (hematocrit <39% in men, <36% in women), on postoperative outcomes. Results: We obtained data for 12,987 patients, of whom 4260 (32.8%) had preoperative anemia. Patients with preoperative anemia experienced higher postoperative major morbidity and mortality rates compared to those without anemia. After adjustment for predefined variables, preoperative anemia was an independent risk factor for postoperative major morbidity (adjusted OR 1.21, ). After adjustment, there was no significant difference in postoperative mortality for patients with or without preoperative anemia (adjusted OR 0.88, ). Conclusion: Preoperative anemia is independently associated with an increased risk of major morbidity in patients undergoing hepatectomy. Therefore, it is crucial to readdress preoperative blood management in anemic patients prior to hepatectomy. Received 21 September 2015; accepted 22 September 2015 Correspondence Allan Tsung, Division of Hepatobiliary and Pancreatic Surgery, Department of Surgery, University of Pittsburgh, 3471 Fifth Avenue, Suite 300, Pittsburgh, PA 15213, USA. Tel: Fax: tsunga@upmc.edu Introduction Anemia is commonly encountered in the preoperative setting when patients are being evaluated for surgery. Preoperative anemia has been shown to be an independent risk factor for increased morbidity and mortality in patients undergoing cardiac surgery. 1 3 In non-cardiac surgery, the prevalence of anemia in the preoperative setting varies widely from as low as 5% in geriatric women having hip surgery to as high as 76% in patients with advanced colorectal cancer undergoing colectomy. 4,5 However, the effect of preoperative anemia on outcomes in non-cardiac surgery has not been firmly established since previous studies This study was presented at the Annual Meeting of the AHPBA, March 2015, Miami, Florida. had limited sample sizes and lacked a uniform definition of anemia. 6 In addition, preoperative anemia is generally viewed as a surrogate marker for overall poorer health of the patient or as a risk factor for postoperative outcomes only due to its association with the higher need for perioperative transfusions. With the development and increased utilization of large national databases such as the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) database, the role of preoperative anemia could be better delineated using a large sample of patients with a unified definition of anemia and ability to adjust for several confounders. Indeed, Musallam et al. studied the role of preoperative anemia in 227,425 patients from the ACS-NSQIP database undergoing major noncardiac surgery in 2008, of whom 30.4% had preoperative anemia. In their analysis, they showed that preoperative anemia

2 256 HPB was associated with increased morbidity and mortality at 30 days after adjustment of several known risk factors including transfusions. 7 However, the same results may not be valid for the subgroup of patients undergoing liver surgery as they constituted <1% of the total sample. In addition, liver surgery is a more complex procedure with more anticipated blood loss when compared to laparoscopic appendectomy or cholecystectomy which were the top procedures in the above analysis. Only a few studies have explored the implications of preoperative anemia and post-operative outcomes specifically in the post-liver resection patients. However, such studies have been only undertaken in specific subgroups of patients such as the elderly and have not adjusted for the major known confounders, especially the use of perioperative transfusions. 8,9 Therefore, in this study, we aimed to identify the prevalence of preoperative anemia in a cohort of patients from the ACS-NSQIP database who underwent liver resection. In addition, we examined whether preoperative anemia is independently associated with higher rates of 30 days morbidity and mortality as this information may trigger changes in the preoperative blood management of patients undergoing liver resection. Methods Data acquisition and patients We analyzed data from the American College of Surgeons National Surgical Quality Improvement program (ACS NSQIP) participant use files for the years We queried hepatectomy procedures by CPT codes and included partial (CPT 47120), right (47130), left (47125) and extended (47122) hepatectomies performed as a primary procedure. Patients undergoing emergency procedures were excluded (n = 143). We also excluded patients with missing data regarding height, weight and gender (n = 90) or preoperative hematocrit (n = 346). Variables Our main independent variable of interest was preoperative anemia, defined as a hematocrit concentration of less than 36% for women and less than 39% for men according to the WHO s criteria. 10 Postoperative outcomes included mortality and morbidity at 30 days. We defined major morbidity as the occurrence of one or more of these serious complications affecting the heart (acute myocardial infarction or cardiac arrest necessitating cardiopulmonary resuscitation), respiratory system (unplanned intubation, ventilator support for >48 h, or pneumonia), central nervous system (cerebrovascular accident with deficits or coma lasting >24 h), kidneys (progressive renal insufficiency requiring replacement therapy), wound (deep incisional surgical site infection, organ or space surgical site infection or wound dehiscence), sepsis or septic shock, pulmonary embolism, or postoperative bleeding as defined by the ACS- NSQIP variable bleeding transfusion, which included any postoperative transfusion given on postoperative day (POD) 1 and beyond. 11 Of note, we defined perioperative transfusion as any transfusion given intraoperatively, in the postoperative recovery room, blood finishing from the operating room, or any blood given on POD 0. Missing data Missing data was almost exclusively limited to preoperative laboratory data, and the proportion of missing data for each variable is listed in Table 1 and Supplementary Table 2. Multiple imputation was performed using predictive mean matching and 5 imputation sets to address this deficiency. 12,13 Summary statistics for the complete cases as well as each imputation set are presented in Supplementary Table 1. Statistical analysis We analyzed several preoperative and perioperative variables between patients with or without anemia with the chi-squared test for categorical variables and the independent samples t-test or Mann Whitney U-test for normal and non-normal continuous variables. The primary outcomes were death or major morbidity at 30 days from the index surgery in the preoperative anemia group compared with the no anemia group. We created separate univariable and multivariable logistic regression models for 30-day mortality and major morbidity. To accommodate model selection within the multiple imputation framework automated stepwise selection procedures (entry threshold 0.15, removal threshold > 0.25) were applied individually to the complete case data as well as each imputation set to generate candidate models. The final multivariable models were constructed by including any predictor exhibiting significant association with the outcome (p < 0.05) in any of the candidate models. Final model estimates were then generated by estimating the final multivariable model in the imputed dataset. We completed all statistical analyses using Stata 13.1 (StataCorp, College Station, TX), and considered two-sided p-values < 0.05 to be statistically significant. Results Of the 13,562 patients who underwent the hepatic procedures of interest, 575 patients met exclusion criteria and 12,987 were included in the analysis. The mean patient age was 59 and 48.4% were male. A total of 4260 (32.8%) patients met criteria for anemia. Of these, 3868 (29.8%) experienced mild anemia (hematocrit >29 <36% for women and >29 <39% in men), while 392 (2%) experienced moderate-severe anemia (hematocrit 29% for both women and men). Both the mild and moderatesevere anemia groups had increased rates of major morbidity and mortality compared to patients without anemia, but since there were relatively few patients with moderate-severe anemia we have decided to combine both groups as one group of patients having anemia in all further analyses.

3 HPB 257 Compared to patients without anemia, those with anemia were more likely to have other associated risk factors including increased age, lower performance status, higher prevalence of cardiovascular, pulmonary, renal, oncological and neurological disorders, higher American Society of Anesthesia (ASA) score, and abnormal preoperative laboratory studies (Table 1 and Supplementary Table 2). Furthermore, patients with anemia were more likely to have a more complex liver resection, longer operative times, longer hospital stay and more likely to receive perioperative transfusion (Table 1). On the other hand, patients with anemia had a lower prevalence of obesity, smoking and recent alcohol intake (Table 1). Major morbidity Of the 12,987 patients in the study, 3390 (26.1%) experienced major morbidity. Major morbidity was significantly higher for patients with anemia (25.6%) than for patients without anemia (17.1%, unadjusted odds ratio [OR unadjusted ] 2.17, 95% CI ). Moreover, compared with patients without anemia, patients with anemia had higher rates of almost all complications with organ space infection and postoperative sepsis the most common serious complication in both the no anemia and anemia groups (10.5% vs. 14.4%, p < 0.001; Table 2). After adjustment for all potential confounders, patients with preoperative anemia were 21% more likely to develop major morbidity within 30 days than those without preoperative anemia (OR adjusted 1.21, 95% CI: ) (Table 3). Anemia was associated with increased risk of major morbidity in patients who did not receive a perioperative transfusion, while anemic patients who did receive a perioperative transfusion were not subject to increased major morbidity. Anemia was significantly associated with an increased risk of major morbidity at 30 days in all age groups, for all surgical procedures, and in patients with and without malignancy (Table 3). Additional factors associated with major morbidity included perioperative transfusion, age and right or extended lobectomy were all also independently associated with an increased risk of major morbidity (p < 0.01, Supplementary Table 3). Thirty-day mortality A total of 279 patients (2.1%) died within 30 days of surgery. Thirty-day mortality was significantly higher for patients with preoperative anemia (3.2%) compared to those without preoperative anemia (1.6%, OR unadjusted 1.98, 95% CI , p < 0.001). However, after adjustment for other risk factors, preoperative anemia was not an independent risk factor for 30- day mortality (Table 3, Supplementary Table 4). The c-statistics for multivariable models ranged from 0.70 to Of note, in our analysis we excluded those patients who were missing preoperative hematocrit levels and gender data. To assess the effect of the patients excluded due to missing data, we evaluated our multivariable models with anemia coded as present or absent for all missing values as a form of sensitivity analysis. We found no significant changes regarding the effect of preoperative anemia on major morbidity and mortality the results when we included the excluded patients in either group for analysis. Discussion In this large multicenter retrospective cohort study, we demonstrate that preoperative anemia is independently associated with an increased risk of 30-day major morbidity in patients undergoing hepatectomy, after adjusting for a number of known confounders. However, preoperative anemia was not independently associated with increased 30-day mortality in our model. Our findings build upon the results of previous studies investigating the role of preoperative anemia on postoperative outcomes. For example, in the largest study to date, Tzeng et al. suggested that anemia (defined as hematocrit <39%) was associated with an increased risk of serious complications in patients undergoing hepatectomies, but their study was limited to elderly patients. 9 Our results build upon these findings and suggest that the negative effects of preoperative anemia persist regardless of age or type of liver surgery. We also found an interesting relationship between preoperative anemia and perioperative transfusion. Blood transfusions can be important in maintaining hemodynamic stability and end organ perfusion during complex surgeries. However, several studies in the different surgical specialties, including hepatopancreaticobiliary surgery, have shown that there may be negative effects of transfusion, including increased length of stay, surgical complications and mortality. 6,14 16 Though anemia and transfusion were independent risk factors for major morbidity in this study, there was no significant additive or multiplicative effect of having preoperative anemia and receiving a transfusion. This is likely because perioperative transfusion restores hematocrit and reduces the risk of complications due to anemia, 7,16,17 but because there are risks inherently associated with transfusion, the overall rate of complications remains the same. The cause of anemia in the general population can be either due to nutritional deficiencies including iron deficiency, chronic disease or of unknown origin. 18 Patients with nutritional and iron deficiency can be easily evaluated and managed preoperatively if encountered. Iron deficiency anemia has been shown to predict worse outcomes after cardiac surgery and treatment of patients with congestive heart failure with parenteral iron improves their cardiac function. 1,19 Several small studies have examined the role of parenteral iron prior to orthopedic surgery and currently a large multicenter randomized controlled trial is currently enrolling in the United Kingdom to assess the role of preoperative intravenous iron to treat anemia in major surgery (PREVENTT). 20,21 Furthermore, iron deficiency anemia is common in patients with malignancies 22 and this warrants special attention as >80% of all patients in our study underwent liver resection for oncological reasons. Treatment of preoperative iron deficiency anemia may be important in this population to avoid

4 258 HPB Table 1 Baseline characteristics of patients No anemia (n [ 8727) Anemia (n [ 4260) p-value General variables Age (Mean SD) 58.1 (13.3) 60.4 (13.4) <0.001 Sex (Male) 4002 (45.9%) 2288 (53.7%) <0.001 Race <0.001 White 6651 (76.1%) 3069 (72.0%) Black 650 (7.4%) 483 (11.3%) Other 1426 (16.3%) 708 (16.6%) Functional status prior <0.001 Independent 8656 (99.2%) 4130 (96.9%) Partially dependent 62 (0.7%) 104 (2.4%) Totally dependent 6 (0.1%) 15 (0.4%) Unknown 3 (<0.1%) 11 (0.3%) BMI (Mean SD) 28.3 (6.3) 27.5 (6.4) <0.001 Alcohol>2 drinks/day within 2 weeks 189 (2.2%) 70 (1.6%) Chronic steroid use 185 (2.1%) 106 (2.5%) 0.18 Extent of resection <0.001 Partial lobectomy 5626 (64.5%) 2514 (59.0%) Extended lobectomy 727 (8.3%) 497 (11.7%) Left lobectomy 838 (9.6%) 454 (10.7%) Right lobectomy 1536 (17.6%) 795 (18.7%) ASA class <0.001 I II 3037 (34.8%) 1023 (24.0%) III 5388 (61.7%) 2929 (68.8%) IV V 302 (3.5%) 308 (7.2%) Operation time (Median, IQR minutes) 214 (152,296) 235 (164,323) <0.001 Perioperative transfusion a 1417 (16.2%) 1405 (33.0%) <0.001 Length of stay (Median IQR) 6 (4,8) 7 (5,10) <0.001 Cardiovascular/Pulmonary variables Smoker within 1 year 1449 (16.6%) 624 (14.6%) Dyspnea at rest or moderate exertion 646 (7.4%) 438 (10.2%) <0.001 Diabetic on oral or insulin 1178 (13.5%) 882 (20.7%) <0.001 Coronary Revascularization (PCI/CABG) 392 (4.5%) 283 (6.6%) <0.001 Severe COPD 235 (2.7%) 156 (3.7%) Hypertension requiring medication 3920 (44.9%) 2244 (52.7%) <0.001 History of revascularization/amputation for PVD 51 (0.6%) 38 (0.9%) CHF (new diagnosis or symptoms within 30 days) 9 (0.1%) 19 (0.4%) <0.001 Recent Angina or Myocardial infarction 28 (0.3%) 24 (0.6%) Currently on dialysis (preoperative) 19 (0.2%) 34 (0.8%) <0.001 Hematological/Oncological variables Preop hematocrit (Mean SD) 41.3 (3.1) 33.8 (3.6) <0.001 Transfusion 1u prbc within 72h (pre-op) 12 (0.1%) 53 (1.2%) <0.001 Malignancy 7239 (82.9%) 3673 (86.2%) <0.001 >10% Weight loss within 6 months 294 (3.4%) 349 (8.2%) <0.001 Chemotherapy within 30 days of Surgery 578 (6.6%) 382 (9.0%) <0.001 Radiotherapy for malignancy within 90 days 61 (0.7%) 57 (1.3%) <0.001 Disseminated cancer 3275 (37.5%) 1736 (40.8%) <0.001

5 HPB 259 Table 1 (continued) No anemia (n [ 8727) Anemia (n [ 4260) p-value Hepatobiliary variables Ascites in previous 30 days 54 (0.6%) 99 (2.3%) <0.001 Esophageal varices in previous 6 months 26 (0.3%) 12 (0.3%) 0.87 Neurological variables History of TIA/CVA 202 (2.3%) 133 (3.1%) Preoperative laboratory variables b Creatinine >1.2 mg/dl 772 (8.8%) 719 (16.9%) <0.001 Total bilirubin >1.0 mg/dl 929 (11.6%) 616 (14.5%) <0.001 AST >40U/L 1860 (21.3%) 1158 (27.2%) <0.001 Alkaline phosphatase >125U/L 1700 (19.5%) 1334 (31.3%) <0.001 Albumin <3 g/dl 112 (1.3%) 501 (11.8%) <0.001 Platelets <150k 1106 (12.7%) 643 (15.1%) <0.001 INR > (1.3%) 174 (4.1%) <0.001 Abbreviations: SD, standard deviation; ASA, American Society of Anesthesiologists; IQR, interquartile range; COPD, chronic obstructive pulmonary disease; PVD, peripheral vascular disease; CHF, congestive heart failure; TIA, transient ischemic attack; CVA, cerebrovascular accident; AST, aspartate aminotransferase, INR, international normalized ratio. a Any transfusion given intraoperatively, in the postoperative recovery room, blood finishing from the operating room, or any blood given on POD 0. b Complete laboratory data including missing presented in Supplementary Table 2. Table 2 Postoperative complications in patients with and without preoperative anemia No anemia (n [ 8727) Anemia (n [ 4260) p-value Any complication 2375 (27.2%) 1756 (41.2%) <0.001 Serious complications 1489 (17.1%) 1091 (25.6%) <0.001 Organ space infection 500 (5.7%) 310 (7.3%) <0.001 Postop sepsis 419 (4.8%) 304 (7.1%) <0.001 Postoperative bleeding 259 (3.0%) 255 (6.0%) <0.001 On ventilator >48h 251 (2.9%) 217 (5.1%) <0.001 Unplanned re-intubation 253 (2.9%) 185 (4.3%) <0.001 Pneumonia 237 (2.7%) 178 (4.2%) <0.001 Postop septic shock 172 (2.0%) 162 (3.8%) <0.001 Renal failure requiring 77 (0.9%) 77 (1.8%) <0.001 dialysis Deep surgical site 76 (0.9%) 60 (1.4%) infection Pulmonary emboli 101 (1.2%) 49 (1.2%) 0.97 Cardiac arrest 60 (0.7%) 48 (1.1%) Wound disruption 62 (0.7%) 40 (0.9%) 0.17 Myocardial infarction 38 (0.4%) 23 (0.5%) 0.41 Cerebrovascular 30 (0.3%) 16 (0.4%) 0.77 accident with deficit Coma >24h 12 (0.1%) 10 (0.2%) 0.21 perioperative transfusions as blood transfusions may have a detrimental effect on cancer specific outcomes after surgery. 15,23 Morbidity and mortality rates after liver surgery have been improving steadily as surgeons have been able to improve modifiable factors that contribute to the development of Table 3 Effect of preoperative anemia on 30 day mortality and major morbidity No anemia (n [ 8727) Anemia (n [ 4260) Mortality N 143 (1.6%) 136 (3.2%) OR unadjusted Reference 1.98 ( ) OR adjusted Reference 0.88 ( ) Major morbidity N 1489 (17.1%) 1091 (25.6%) OR unadjusted Reference 1.67 ( ) OR adjusted Reference 1.21 ( ) Strata OR adjusted major morbidity Age <65 Reference 1.18 ( ) 65 Reference 1.26 ( ) Perioperative transfusion Yes Reference 0.99 ( ) No Reference 1.32 ( ) Extent of resection Partial lobectomy Reference 1.19 ( ) Right/Left/Extended Reference 1.21 ( ) Malignancy No Reference 1.37 ( ) Yes Reference 1.18 ( ) complication after liver surgery. Tzeng et al. identified four categories of risk factors for complication after liver resection: surgical experience and volume, the culture of safety at a given

6 260 HPB hospital, extent of resection, and patient factors. 9 In recent years, centralization of major liver resections to high volume and teaching centers have contributed to improved patient outcomes Meanwhile, there may be opportunities for the surgeon to carefully consider he last two categories during preoperative planning. For instance, the surgeon s judgmentin using the appropriate extent of resection is important because larger resections are associated with higher rates of postoperative complications. However,with the improvements in surgical technique such as minimally invasive options and advances in anesthetic and postoperative care have allowed for complex procedures to be conducted without increased morbidity. In addition to choosing the least invasive procedure appropriate for the patient s condition, surgeons have the option to optimize preoperative risk factors to reduce the incidence of complications. Our results strongly suggest that more emphasis may need to be placed on identifying and managing anemia preoperatively to improve outcomes and reduce requirements for perioperative transfusion. However, managing anemia may be more successful for some etiologies than for others. For example, in many patients, iron-deficiency anemia can be treated via supplementation in the outpatient setting, while anemia associated with an underlying malignancy may be markedly more difficult to treat before surgery. In patients who can be treated, complete resolution of anemia may not occur before the desired date of surgical intervention. However, there is a lower risk of complications in patients with mild vs. moderate-severe anemia, 7 so initiation of treatment before surgery may be beneficial. This study has some limitations. First, the NSQIP database only includes deaths and complications that occur within 30 days of surgical intervention, so we could not determine the effects of anemia on long-term complications. Second, few data are available for complications specific to hepatectomy, which may underestimate postoperative morbidity in this population. Third, it is not possible to identify patients who received less than four units of blood preoperatively using NSQIP data. Consequently, our analysis may have included a small number of patients who received less than 4 units of blood preoperatively although our intent was to exclude all preoperative transfusions. In addition, information about the etiology and time course of anemia was not available, though this information may be relevant to postoperative outcomes. Despite these limitations, the NSQIP database is a large and comprehensive sample of surgical experience across the United States and, because it includes so many perioperative variables, is an excellent tool for assessing the effects of preoperative factors on short-term outcomes. In conclusion, our findings suggest that anemia is a serious condition that negatively impacts outcomes after liver resection. Whenever possible, efforts to identify and treat anemia prior to elective liver resection may be useful in maintaining physiological hemoglobin levels throughout the perioperative period. In addition to reducing the cost of care, successful management of preoperative anemia may ultimately result in better patient outcomes and improved quality of life. ACS-NSQIP disclaimer The American College of Surgeons National Quality Improvement Program and the hospitals participating in the ACS NSQIP represent the source of the data used herein; these institutions have not verified and are not responsible for the statistical validity of the data analysis or the conclusions derived by the authors. Funding sources Samer Tohme ST is funded by Americas Hepato-Pancreato-Biliary Association Research Award (S.T.), the National Cancer Institute Grant Number T32CA Conflicts of interest None to declare. References 1. Karkouti K, Wijeysundera DN, Beattie WS. (2008 Jan 29) Reducing bleeding in cardiac surgery I. Risk associated with preoperative anemia in cardiac surgery: a multicenter cohort study. Circulation 117: PubMed PMID: Kulier A, Levin J, Moser R, Rumpold-Seitlinger G, Tudor IC, Snyder- Ramos SA et al. (2007 Jul 31) Impact of preoperative anemia on outcome in patients undergoing coronary artery bypass graft surgery. Circulation 116: PubMed PMID: van Straten AH, Hamad MA, van Zundert AJ, Martens EJ, Schonberger JP, de Wolf AM. (2009 Jul 14) Preoperative hemoglobin level as a predictor of survival after coronary artery bypass grafting: a comparison with the matched general population. Circulation 120: PubMed PMID: Cappell MS, Goldberg ES. (1992 Apr) The relationship between the clinical presentation and spread of colon cancer in 315 consecutive patients. A significant trend of earlier cancer detection from 1982 through 1988 at a university hospital. J Clin Gastroenterol 14: PubMed PMID: Haljamae H, Stefansson T, Wickstrom I. (1982 Aug) Preanesthetic evaluation of the female geriatric patient with hip fracture. Acta Anaesthesiol Scand 26: PubMed PMID: Shander A, Knight K, Thurer R, Adamson J, Spence R. (2004 Apr 5) Prevalence and outcomes of anemia in surgery: a systematic review of the literature. Am J Med 116(Suppl. 7A):58S 69S. PubMed PMID: Musallam KM, Tamim HM, Richards T, Spahn DR, Rosendaal FR, Habbal A et al. (2011 Oct 15) Preoperative anaemia and postoperative outcomes in non-cardiac surgery: a retrospective cohort study. Lancet 378: PubMed PMID: Shubert CR, Habermann EB, Truty MJ, Thomsen KM, Kendrick ML, Nagorney DM. (2014 Nov) Defining perioperative risk after hepatectomy based on diagnosis and extent of resection. J Gastrointest Surg Off J Soc Surg Aliment Tract 18: PubMed PMID: Tzeng CW, Cooper AB, Vauthey JN, Curley SA, Aloia TA. (2014 May) Predictors of morbidity and mortality after hepatectomy in elderly patients: analysis of 7621 NSQIP patients. HPB Off J Int Hepato Pancreato Biliary Assoc 16: PubMed PMID: Pubmed Central PMCID:

7 HPB Nutritional anaemias. (1968) Report of a WHO scientific group. World Health Organ Tech Rep Ser 405:5 37. PubMed PMID: Ghaferi AA, Birkmeyer JD, Dimick JB. (2009 Oct 1) Variation in hospital mortality associated with inpatient surgery. N Engl J Med 361: PubMed PMID: Rubin DB, Schenker N. (1991 Apr) Multiple imputation in health-care databases: an overview and some applications. Stat Med 10: PubMed PMID: Wood AM, White IR, Royston P. (2008 Jul 30) How should variable selection be performed with multiply imputed data? Stat Med 27: PubMed PMID: Ferraris VA, Davenport DL, Saha SP, Austin PC, Zwischenberger JB. (2012 Jan) Surgical outcomes and transfusion of minimal amounts of blood in the operating room. Arch Surg 147: PubMed PMID: Ejaz A, Spolverato G, Kim Y, Margonis GA, Gupta R, Amini N et al. (2015 May) Impact of blood transfusions and transfusion practices on long-term outcome following hepatopancreaticobiliary surgery. J Gastrointest Surg Off J Soc Surg Aliment Tract 19: PubMed PMID: Clevenger B, Richards T. (2015 Jan) Pre-operative anaemia. Anaesthesia 70(Suppl. 1), 20-8, e6 8. PubMed PMID: Spahn DR. (2010 Aug) Anemia and patient blood management in hip and knee surgery: a systematic review of the literature. Anesthesiology 113: PubMed PMID: Guralnik JM, Eisenstaedt RS, Ferrucci L, Klein HG, Woodman RC. (2004 Oct 15) Prevalence of anemia in persons 65 years and older in the United States: evidence for a high rate of unexplained anemia. Blood 104: PubMed PMID: Anker SD, Comin Colet J, Filippatos G, Willenheimer R, Dickstein K, Drexler H et al. (2009 Dec 17) Ferric carboxymaltose in patients with heart failure and iron deficiency. N Engl J Med 361: PubMed PMID: Gonzalez-Porras JR, Colado E, Conde MP, Lopez T, Nieto MJ, Corral M. (2009 Feb) An individualized pre-operative blood saving protocol can increase pre-operative haemoglobin levels and reduce the need for transfusion in elective total hip or knee arthroplasty. Transfus Med 19: PubMed PMID: Garcia-Erce JA, Cuenca J, Martinez F, Cardona R, Perez-Serrano L, Munoz M. (2006 Oct) Perioperative intravenous iron preserves iron stores and may hasten the recovery from post-operative anaemia after knee replacement surgery. Transfus Med16: PubMedPMID: Dicato M, Plawny L, Diederich M. (2010 Oct) Anemia in cancer. Ann Oncol Off J Eur Soc Med Oncol/ESMO 21(Suppl. 7):vii167 vii172. PubMed PMID: Amato A, Pescatori M. (2006) Perioperative blood transfusions for the recurrence of colorectal cancer. Cochrane Database Syst Rev, CD PubMed PMID: Choti MA, Bowman HM, Pitt HA, Sosa JA, Sitzmann JV, Cameron JL et al. (1998 Jan Feb) Should hepatic resections be performed at highvolume referral centers? J Gastrointest Surg Off J Soc Surg Aliment Tract 2: PubMed PMID: Glasgow RE, Showstack JA, Katz PP, Corvera CU, Warren RS, Mulvihill SJ. (1999 Jan) The relationship between hospital volume and outcomes of hepatic resection for hepatocellular carcinoma. Arch Surg 134: PubMed PMID: Dimick JB, Wainess RM, Cowan JA, Upchurch, GR, Jr., Knol JA, Colletti LM. (2004 Jul) National trends in the use and outcomes of hepatic resection. J Am Coll Surg 199: PubMed PMID: Appendix A. Supplementary data Supplementary data related to this article can be found at /j.hpb