A significant increase was observed in operative complications over time during the study period, consisting mostly of an increase in bleeding and transfusion. Operative mortality remained less than 1% in 2010 and 2011.
A significant increase was observed in bleeding requiring transfusion during the study period. The cause for this change remains unclear.
Predicted risk is shown for complications based on significant preoperative clinical and pathologic variables. Points for each variable are assigned based on the clinical variable, and the point summary corresponds to the approximate risk for an operative complication. INR indicates international normalized ratio. To convert albumin level to grams per liter, multiply by 10; to convert bilirubin level to micromoles per liter, multiply by 17.104.
Low preoperative albumin level significantly increases the risk for operative mortality, and older age is a relative contraindication for elective splenectomy. To convert albumin level to grams per liter, multiply by 10.
Bagrodia N, Button AM, Spanheimer PM, Belding-Schmitt ME, Rosenstein LJ, Mezhir JJ. Morbidity and Mortality Following Elective Splenectomy for Benign and Malignant Hematologic ConditionsAnalysis of the American College of Surgeons National Surgical Quality Improvement Program Data. JAMA Surg. 2014;149(10):1022-1029. doi:10.1001/jamasurg.2014.285
Splenectomy is a commonly performed operation; however, data from large series regarding operative outcomes to help guide decision making and informed consent are lacking.
To evaluate clinical and pathologic variables associated with morbidity and mortality following elective splenectomy for benign and malignant hematologic conditions in the United States.
Design, Setting, and Participants
A review of the American College of Surgeons National Surgical Quality Improvement Program data for elective splenectomy between January 1, 2005, and December 31, 2011, was performed, and 1715 eligible individuals were identified.
Elective splenectomy for hematologic conditions.
Main Outcomes and Measures
Complications and operative mortality were evaluated for the entire cohort and compared between patients with benign vs malignant diseases. Multivariable logistic regression was used to evaluate factors predictive of operative complications and death.
Splenectomy was performed in 1344 patients (78.4%) for benign disease and in 371 patients (21.6%) for malignant disease. Two hundred ninety-one patients (17.0%) had a complication, and operative mortality occurred in 27 patients (mortality rate, 1.6%). Patients treated for malignant disease had a higher rate of overall complications (27.2%) compared with patients treated for benign disease (14.1%) (P < .001). Several variables were independent predictors of complications, including malignant disease (vs benign) (Odds Ratio [OR], 1.86; 95% CI, 1.23-2.80; P = .003), independent performance status (vs dependent) (OR, 0.33; 95% CI, 0.07-1.52; P = .02), and increasing albumin level (OR, 0.75; 95% CI, 0.66-0.86; P < .001). Increasing age (OR, 1.03; 95% CI, 1.00-1.06; P = .05) was an independent predictor of mortality while increasing albumin level (OR, 0.63; 95% CI, 0.46-0.86; P = .003) predicted lower risk of operative death. From these data, a patient older than 60 years with a low preoperative albumin level has a predicted probability for operative death as high as 10.0%.
Conclusions and Relevance
Preoperative performance and nutritional status are significant risk factors for complications and mortality following elective splenectomy. Although operative mortality continues to decrease over time, specific preoperative variables may help with patient selection before elective splenectomy for certain patients.
Splenectomy is a commonly performed operation for various conditions, including trauma and benign and malignant hematologic disorders.1,2 The operation is frequently performed laparoscopically; however, some patients still undergo open splenectomy depending on patient-specific factors and surgeon preference.3 Many hematologic conditions such as hemolytic anemia and idiopathic thrombocytopenic purpura are treated with splenectomy following failure of medical management.4,5 Malignant conditions such as leukemia and lymphoma are treated selectively with splenectomy when the spleen is primarily involved, with the potential to improve survival over that of chemotherapy alone.6 In addition, splenectomy may be performed palliatively to help control pain or to decrease transfusion needs in certain patients. Not all patients may benefit from removal of the spleen irrespective of the approach; therefore, careful consideration of the risks and benefits of the operation need to be considered when offering the procedure to treat a given hematologic disease.7
Splenectomy has been shown to be a safe and effective treatment option for select hematologic conditions, with low complication rates and mortality.8,9 However, operative mortality from splenectomy for some hematologic disorders has been reported to be as high as 6.3%.9,10 It has also been demonstrated in large retrospective investigations that laparoscopic splenectomy, when feasible, is associated with fewer complications, less blood transfusion, shorter hospital stay, and faster return to work.11 However, data are limited on the operative morbidity and mortality associated with elective splenectomy from large multi-institutional series.12,13 Many studies published are from single institutions and do not compare the results between patients treated for benign vs malignant hematologic conditions, who may have different outcomes for numerous reasons.
The objective of this study was to evaluate the variables predictive of complications and mortality following elective splenectomy using a large national data set. This analysis will provide information to guide clinical decision making and informed consent for this commonly performed procedure.
The American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) data set was used to identify patients who underwent elective splenectomy for benign and malignant hematologic diseases between January 1, 2005, and December 31, 2011. The ACS NSQIP collects data on patients being treated with surgical procedures in the United States from university and private medical centers. Patient clinicopathologic and treatment-related variables are collected in addition to 30-day outcomes, including operative complications and mortality. The ACS NSQIP now collects 252 case variables and has been the focus of publications focused on operative outcomes and quality assessment.14- 16 The University of Iowa Institutional Review Board deemed this study to be exempt from review due to the retrospective nature of the analysis of deidentified data.
Benign hematologic conditions included hemolytic anemia, idiopathic thrombocytopenic purpura, thrombotic thrombocytopenic purpura, hereditary spherocytosis, polycythemia vera, thrombocytopenia, and Evans syndrome. Malignant hematologic conditions included leukemia and lymphoma. Excluded from analysis were patients who had a traumatic splenic injury, patients who received a resection for splenomegaly or hypersplenism, patients who underwent nonelective splenectomy (because of trauma or accidental intraoperative injury), and patients who had a splenectomy performed in addition to another major procedure (eg, cholecystectomy) or a multivisceral resection (eg, colectomy or pancreatectomy). Patients without an identifiable diagnosis were also excluded.
Postoperative complications were recorded. These included renal failure, perioperative transfusion, deep venous thrombosis or pulmonary embolism, pulmonary complications (reintubation or failure to wean from the ventilator), and infectious complications (wound infection, pneumonia, urinary tract infection, or sepsis), as well as operative death within 30 days.
The variables were analyzed with comparative statistics using t test and χ2 test where appropriate. Univariate logistic regression models were fit for each of the predictor variables, and the results were reported for all of the operative complication outcome variables. For each end point, multivariable logistic models were built with the significantly associated predictors using stepwise selection. Estimated associations for the univariate and multivariable models are reported as odds ratios (95% CIs). A nomogram for risk estimation for operative complications was constructed to demonstrate the relative importance of each predictor in the multivariable logistic model. An effect plot for operative mortality was created using significant demographic and pathologic variables from the multivariable analysis to determine patients at highest risk for death following splenectomy. Statistical tests were 2-sided and were assessed for significance at the 5% level. Analyses were performed with a statistical software package (SAS 9.3; SAS Institute Inc).
In total, 4907 patients were identified by the ACS NSQIP who had a splenectomy between 2005 and 2011, and 1715 patients met criteria for analysis. Clinicopathologic variables for the entire cohort are listed in Table 1. Quiz Ref IDOverall, 1344 patients (78.4%) were treated for benign hematologic disorders, including 988 (57.6%) with idiopathic thrombocytopenic purpura and 153 (8.9%) with hemolytic anemia. In total, 371 patients (21.6%) were treated for malignant hematologic conditions, including 59 (3.4%) with leukemia and 312 (18.2%) with lymphoma. Eighty-four patients (4.9%) had impaired functional status before surgery.
Treatment-related variables and operative outcomes are listed in Table 1. Quiz Ref IDIn total, 1258 patients (73.4%) were treated with laparoscopic splenectomy. Twenty-seven patients died (overall operative mortality rate, 1.6%), and 291 patients (17.0%) experienced at least 1 complication. Despite a slight increase in morbidity over time, a significant trend of decreased mortality was observed during the study period (Figure 1). No corresponding significant change over time was seen in the percentage of patients treated for malignant disease or in the incidence of infectious complications. Also, no significant change was observed in the percentage of patients who received laparoscopic splenectomy over time: the percentages were 74.9% (in 2005-2007), 71.8% (in 2008-2009), and 73.5% (in 2001-2011).
The percentage of patients undergoing transfusion remained less than 2% from 2005 to 2008 and significantly increased in 2010 and 2011 (Figure 2). No significant change was observed in the percentage of patients requiring preoperative transfusion or in the preoperative international normalized ratio (INR) over time. Among 36 patients who had a preoperative transfusion, only 11 required an additional perioperative transfusion. To determine the effect of the preoperative INR on operative transfusion, the mean preoperative INRs were compared among patients who underwent a transfusion and patients who did not receive a transfusion. Patients who underwent a transfusion had a higher mean (SD) preoperative INR of 1.13 (0.17) compared with 1.07 (0.22) in patients who did not receive a transfusion (P = .02).
Table 2 summarizes the clinical and treatment-related variables of patients treated for malignant vs benign hematologic conditions. Quiz Ref IDPatients with malignant disease more often had preoperative weight loss and dependent functional status. A higher percentage of patients underwent a laparoscopic procedure for benign disease (P < .001) and had shorter operative time and hospital stay. The complication rate was significantly higher for patients with malignant disease (27.2% vs 14.1%, P < .001). This was mostly reflected in infectious complications and bleeding requiring transfusion. Operative mortality was not significantly different, at 1.6% for both groups (P = .94).
The end points evaluated were those with an adequate number of events for statistical evaluation among 822 patients with all available data points (overall and infectious complications and operative mortality). Among the variables associated with overall complications were age, diabetes mellitus, cardiovascular disease, chronic corticosteroid use, impaired functional status, procedure type (laparoscopic vs open), operative time, malignant pathology, and preoperative albumin and bilirubin levels and INR (Table 3). The variables associated with operative mortality included age, race/ethnicity, cardiovascular disease, impaired functional status, and preoperative albumin level.
Malignant disease remained an independent risk factor for complications (Table 4). Fitness and nutritional status were important predictors of complications as evidenced by dependent functional status, decreasing preoperative albumin level, and increasing preoperative bilirubin level and INR. Factors associated with infectious complications on multivariable analysis included decreasing preoperative albumin level, increasing preoperative bilirubin level and INR, dependent functional status, and longer operative time.
Quiz Ref IDFactors that remained independent predictors of mortality on multivariable analysis included decreasing preoperative albumin level and increasing age. After adjustment for other significant variables, the multivariable analysis revealed that an increase in preoperative albumin level by 0.5 g/dL decreased the risk for operative mortality by a factor of 0.63 (P = .003) (to convert albumin level to grams per liter, multiply by 10).
To illustrate the effect of preoperative clinical variables on operative complications and mortality, a nomogram was generated using the results from the multivariable analysis. As shown in Figure 3, a patient with benign disease who exhibits dependent functional status (30 points), a preoperative albumin level of 2.0 g/dL or less (85 points), and an elevated preoperative INR (45 points) has a predicted complication rate of almost 70% (total point score, 160). A patient with malignant disease who is independent but has an elevated preoperative INR and a low preoperative albumin level has a predicted complication rate exceeding 80%.
An effect plot was generated to illustrate the weighted risk for operative mortality following elective splenectomy based on significant variables from the multivariable analysis. Increasing age and decreasing preoperative albumin level are predictive of operative mortality (Figure 4).
Quiz Ref IDThis study demonstrates that elective splenectomy for the treatment of malignant conditions has a higher complication rate, but an equivalent mortality rate, compared with splenectomy for the treatment of benign conditions. This evaluation allowed us to determine multiple preoperative clinical variables that independently predict operative complications, including malignant diagnosis, smoking, diabetes mellitus, poor functional status, low preoperative albumin level, and elevated preoperative bilirubin level and INR. The variables that predict mortality on multivariable analysis include increased age and preoperative malnutrition as evidenced by albumin level. This study also shows that, despite an increase in operative complications and transfusions over time, mortality decreased from 3.0% in 2005 to 0.9% in 2011. Although this was a nonsignificant trend statistically over time, it demonstrates a lower risk than in previous studies for benign diseases4 and malignant diseases.7,9,13
The laparoscopic approach to splenectomy for benign and malignant conditions has been shown to be safe and effective.3,5,11,12,17 A previous study11 using the ACS NSQIP data comparing laparoscopic with open splenectomy among 1781 patients in 2008 and 2009 demonstrated that laparoscopic splenectomy was associated with fewer complications, less blood transfusion, and lower operative mortality. The authors accounted for splenectomy indications using multivariable analysis, but it is still possible that the open and laparoscopic groups were biased based on patient selection for the procedure performed (eg, splenomegaly).11 Similarly, patients herein having benign disease more often underwent laparoscopic splenectomy (82.4%) compared with those having malignant disease (40.4%) (P < .001). This may account for some of the differences in outcomes between the 2 groups, although the multivariable analysis did not demonstrate operative approach as an independent predictor of complications or mortality. Furthermore, multiple factors may lead to an open approach for patients with malignant disease such as splenomegaly, coagulopathy, and the potential need to dissect a large-diameter splenic vein close to the superior mesenteric vein because of risk for portal vein thrombosis.18
The overall complication rate in this series was 17.0%, and patients having malignant disease (27.2%) had a higher rate of complications compared with patients having benign disease (14.1%). Of these, infectious complications and bleeding with the need for transfusion predominated. The overall rate of infectious complications was 8.4% (12.7% in patients with malignant disease and 7.2% in patients with benign disease). Although numerous studies have reported splenectomy outcomes, it is difficult to determine specific complication rates for patients with malignant vs benign diseases because the investigations have focused on operative approach and combined these patients in the analyses. A recent large series of patients treated for malignancy revealed that infection was the most common postoperative complication (15.0% overall).9 Other large multi-institutional studies8,17,19 report postoperative infection rates ranging from 1.0% to 6.3% for benign and malignant conditions.
The percentage of patients who had bleeding requiring transfusion increased over time. It is unclear what is responsible for this. Although malignant disease was associated with an increased risk for bleeding and transfusion, the percentage of patients who were treated for malignant disease or who underwent laparoscopic splenectomy did not change over time in this study. Patients who underwent a transfusion had a higher mean preoperative INR compared with patients who did not undergo a transfusion; however, the preoperative INR did not change during the time frame of the study. Musallam and colleagues11 found the rate of transfusion following splenectomy to be 29.4% (11.2% for laparoscopic splenectomy vs 47% for open splenectomy). Open splenectomy was a significant risk factor for transfusion even after controlling for transfusion propensity based on multiple clinical variables. The high rates of transfusion in the study by Musallam et al may be explained by the fact that patients were included who had nonelective procedures, multivisceral resection, and traumatic intraoperative injury.
The incidence of thrombotic complications (deep venous thrombosis and pulmonary embolism) was equivalent between patients with benign vs malignant conditions and was 2.4% for the entire cohort. In a population-based cohort study20 from Denmark, 3812 patients were evaluated following splenectomy for thrombotic complications. Venous thromboembolism occurred in 71 patients (1.9%) within 90 days after splenectomy, which was higher than the comparison groups of the general population (0.06%) or patients following appendectomy (0.6%). The incidence of venous thromboembolism in their study is similar to that in other series reporting rates of this complication following splenectomy ranging from 1.8%8 to 3.3%.11 Unfortunately, the ACS NSQIP does not collect data on DVT prophylaxis administration, so we could not evaluate the effect of this practice on operative complications. Also, the ACS NSQIP database does not provide long-term follow-up data, so the true incidence of these complications over time is unknown. This database also does not collect information on portal vein thrombus, which is an important and recognized complication from this operation, especially in patients with a large-diameter splenic vein.18
Numerous studies have shown that mortality from pancreatectomy and hepatectomy is as low as 1% to 2% in high-volume centers. It is remarkable that the rate for death from elective splenectomy in the present large study is 1.6%, making it a procedure of comparable risk. The mortality rate in our study was 3.0% in 2005 and was down to 0.9% in 2011. This rate is lower than rates reported for patients with myeloproliferative disease, splenomegaly, and other malignancies, in whom mortality ranges from 1.8% to 9.0%.7- 9,13 Not all of these diagnoses were included in the present study. This study also does not have information to stratify across different types of lymphoma and leukemia, as well as whether they were acute or chronic in nature. These are variables that may affect survival following this operation and should be considered when selecting patients for treatment with splenectomy.
From these data, measures may be taken to reduce complications from splenectomy. As mentioned above, the most common potentially preventable complications observed were infectious complications, venous thromboembolism, and bleeding and transfusion. Overwhelming postsplenectomy sepsis is a rare complication following elective splenectomy in the immediate postoperative period and overall remains uncommon because of the routine use of vaccinations.21 Several interventions may reduce postoperative infectious complications in elective surgical procedures, including antibiotic prophylaxis, intraoperative temperature control, and early catheter removal.22,23 Although these variables were unavailable for analysis in this study, they may be applied to patients undergoing splenectomy. Prophylaxis for venous thromboembolism during surgery and in the early postoperative period is critical in all patients and especially those with malignant disease. Patients undergoing splenectomy often have thrombocytopenia, which may reduce the perceived need for thromboprophylaxis. Despite thrombocytopenia, chemical prophylaxis in these patients is warranted.24 Similar to our findings, preoperative functional status and nutritional status have been previously shown to be important predictors of outcomes for splenectomy.9 It is frequently impossible to improve functional status, although performance often correlates with nutritional status, and the condition of patients should perhaps be optimized before surgery if timing allows. Although the preoperative INR can be corrected with plasma transfusions or vitamin K treatment (phytonadione), an elevated preoperative INR may mirror the severity of illness in a patient and is a significant predictor of complications.
Risk factors for transfusion such as preoperative anemia, complexity of the operation being performed, and coagulopathy are well documented in general surgical procedures. Transfusion of 1 U of blood has been shown to be associated with a significant increase in the risk for overall complications, infectious complications, and mortality following general surgical procedures, including splenectomy.25 Judicious transfusion practices may help reduce the incidence of transfusion and the associated consequences26; however, transfusion remains a lifesaving practice in patients who have significant hemorrhage.
This study has limitations. The ASC NSQIP database represents a sampling of procedures; despite being a large data set, it does not record consecutive operations performed during the period analyzed. Also, the ACS NSQIP is an administrative data set that may have inaccuracies in the reporting of data. As mentioned above, some diagnoses are not represented, which may indicate a lack of generalizability of the data to all patients. Furthermore, outcomes in the ACS NSQIP are limited to 30 days following surgery, so certain complications are not comprehensively evaluable such as venous thromboembolism and postsplenectomy sepsis. Most important, emergency splenectomy procedures, including patients who had trauma, patients who underwent treatment of splenomegaly, and patients who required additional procedures or multivisceral resection, were excluded from analysis. These operations are expected to have a higher complication rate and possibly mortality and should be considered distinct from these outcomes data.
These data provide a benchmark for complications and mortality from elective splenectomy procedures for patients with benign and malignant hematologic disorders. The differences in complications highlight the need to evaluate these patients separately when reporting outcomes from this procedure because combing them may not provide the appropriate information to help with patient selection. The mortality rate overall was low for patients with malignant and benign diseases; however, among patients with certain preoperative risk factors, morbidity can be as high as 80.0% and mortality as high as 10.0% (Figure 3). These data may be used to help guide decision making when considering splenectomy for patients with benign and malignant hematologic disorders.
Accepted for Publication: January 6, 2014.
Corresponding Author: James J. Mezhir, MD, Division of Surgical Oncology and Endocrine Surgery, Department of Surgery, University of Iowa Hospitals and Clinics, 200 Hawkins Dr, 4642 JCP, Iowa City, IA 52242 (firstname.lastname@example.org).
Published Online: August 20, 2014. doi:10.1001/jamasurg.2014.285.
Author Contributions: Dr Mezhir had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Mezhir.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Bagrodia, Button, Belding-Schmitt, Mezhir.
Critical revision of the manuscript for important intellectual content: Button, Spanheimer, Rosenstein, Mezhir.
Statistical analysis: Button, Spanheimer.
Administrative, technical, or material support: Spanheimer.
Study supervision: Mezhir.
Conflict of Interest Disclosures: None reported.
Previous Presentations: This study was presented in part at 54th Annual Meeting of The Society for Surgery of the Alimentary Tract; May 21, 2013; Orlando, Florida; and at the American College of Surgeons National Surgical Quality Improvement Program 2013 National Conference; July 15, 2013; San Diego, California.