ORIGINAL ARTICLE
Postoperative Complications Reduce Adjuvant Chemotherapy Use in Resectable Pancreatic Cancer Ryan P. Merkow, MD, MS,∗ †‡ Karl Y. Bilimoria, MD, MS,∗ † James S. Tomlinson, MD, PhD,§ Jennifer L. Paruch, MD,∗ ‡ Jason B. Fleming, MD,¶ Mark S. Talamonti, MD,‡|| Clifford Y. Ko, MD, MS, MSHS,∗ § and David J. Bentrem, MD, MS†∗∗ Objective: To assess the impact of postoperative complications on the receipt of adjuvant chemotherapy. Background: Randomized trials have demonstrated that adjuvant chemotherapy is associated with improved long-term survival. However, pancreatic surgery is associated with significant morbidity and the degree to which complications limit subsequent treatment options is unknown. Methods: Patients from the American College of Surgeons National Surgical Quality Improvement Program and the National Cancer Data Base who underwent pancreatic resection for cancer were linked (2006–2009). The associations between complications and adjuvant chemotherapy use or treatment delay (≥70 days from surgery) were assessed using multivariable regression methods. Results: From 149 hospitals, 2047 patients underwent resection for stage I-III pancreatic adenocarcinoma of which 23.2% had at least 1 serious complication. Overall adjuvant chemotherapy receipt was 57.7%: 61.8% among patients not experiencing any complication and 43.6% among those who had a serious complication. Serious complications increased the likelihood of not receiving adjuvant therapy over twofold [odds ratio (OR) = 2.20, 95% confidence interval (CI): 1.73–2.80]. Specific complications associated with adjuvant chemotherapy omission were reintubation (OR = 7.79, 95% CI: 3.59–16.87), prolonged ventilation (OR = 5.92, 95% CI: 3.23–10.86), pneumonia (OR = 2.83, 95% CI: 1.63–4.90), sepsis/shock (OR = 2.76, 95% CI: 2.02–3.76), organ space/deep surgical site infection (OR = 2.19, 95% CI: 1.53–3.13), venous thromboembolism (OR = 1.92, 95% CI: 1.08–3.43), and urinary tract infection (OR = 1.61, 95% CI: 1.02–2.54). Serious complications also doubled the likelihood of delaying adjuvant treatment administration (OR = 2.08, 95% CI: 1.42–3.05). Sensitivity analysis in a younger, healthier patient cohort demonstrated similar associations. Conclusions: Postoperative complications are common following pancreatic surgery and are associated with adjuvant chemotherapy omission and treatment delays.
Keywords: ACS NSQIP, adjuvant therapy, complications, NCDB, pancreatic cancer, surgery (Ann Surg 2014;260:372–377)
I
n 2013, approximately 45 220 new cases of pancreatic cancer will be diagnosed and 38 460 people will succumb to the disease.1 By 2020, it is estimated that pancreatic cancer will become the second most common cause of cancer death.2 Although mortality after surgery has decreased markedly in recent years, complications remain common. More than 30% of patients experience at least 1 complication, and serious complications occur in up to 20% of patients.3,4 Although surgical resection is the primary component of multimodality therapy, pancreatic cancer is a biologically aggressive disease. To date, randomized trial data has demonstrated that adjuvant chemotherapy is associated with improved long-term survival when combined with surgical resection.5–8 Currently, national guidelines recommend adjuvant chemotherapy administration in the postoperative setting.9 Nevertheless, evidence suggests that more than 30% of patients who are eligible for systemic chemotherapy are never treated.10 Given pancreatic resection is associated with significant morbidity, the degree to which complications limit subsequent treatment options is unknown. If a strong relationship exists between complications and adjuvant therapy omission, theoretically, administering chemotherapy upfront before surgery could potentially increase the number of patients who would ultimately benefit from its effects.11,12 Our objectives were to determine the association of postoperative complications with adjuvant therapy receipt and, among patients who receive adjuvant therapy, assess the impact of complications on time to treatment.
METHODS Data Sources From the ∗ Division of Research and Optimal Patient Care, American College of Surgeons, Chicago, IL; †Surgical Outcomes and Quality Improvement Center and the Northwestern Institute for Comparative Effectiveness Research (NICER) in Oncology, Department of Surgery, Northwestern University Feinberg School of Medicine, Chicago, IL; ‡Department of Surgery, University of Chicago Pritzker School of Medicine, Chicago, IL; §Department of Surgery, University of California, Los Angeles and VA Greater Los Angeles Healthcare System, Los Angeles, CA; ¶Department of Surgical Oncology, MD Anderson Cancer Center, Houston, TX; ||Department of Surgery, Northshore University Health System, Evanston, IL; and ∗∗ Department of Surgery, Jesse Brown Veteran Affairs Medical Center, Chicago, IL. Disclosure: The authors have no conflicts of interest to disclose. This study was supported by a grant from the American Cancer Society (No. 280521) to R.P.M. and K.Y.B. J.L.P. is supported by a grant from Genetech. Presented at the 66th Society of Surgical Oncology Annual Cancer Symposium, March 8, 2013, National Harbor, MD. Reprints: Ryan P. Merkow, MD, MS, Division of Research and Optimal Patient Care, American College of Surgeons, 633 N Saint Clair Street, 22nd Floor, Chicago, IL 60611. E-mail: [email protected]. C 2013 by Lippincott Williams & Wilkins Copyright ISSN: 0003-4932/13/26002-0372 DOI: 10.1097/SLA.0000000000000378
372 | www.annalsofsurgery.com
The American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) and the National Cancer Data Base (NCDB) were linked to evaluate the effects of postoperative complications on adjuvant therapy use after pancreatic cancer surgery. ACS NSQIP contains detailed patient, demographic, comorbidity and complication data,13,14 while the NCDB collects cancer-specific tumor and treatment information.15 Because a unique patient identifier common to each data set does not exist, patients from ACS NSQIP were linked to the NCDB on the basis of a probabilistic matching algorithm using the following variables: American Hospital Association number, patient age, sex, and date of operation. The final match rate was 71%, which is comparable to the linkage of NSQIP with other data sources without the use of direct patient identifiers.16 The developmental history and current details of ACS NSQIP, including sampling strategy, data abstraction procedures, variables collected, complications assessed, and structure are described elsewhere.17–20 In brief, hospitals collect standardized and audited data by trained surgical clinical reviewers using standard ACS NSQIP Annals of Surgery r Volume 260, Number 2, August 2014
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Annals of Surgery r Volume 260, Number 2, August 2014
tools and definitions.21 Patients are followed for postoperative complications for 30 days after the index operation irrespective of whether the patient is an inpatient, has been discharged to his/her home or another facility, or has been readmitted to another hospital. The NCDB is a joint program of the American College of Surgeons, Commission on Cancer, and the American Cancer Society.22 NCDB is the largest cancer registry in the world, capturing more than 70% of all newly diagnosed malignancies in the United States from more than 1500 hospitals.23 Trained and audited cancer registrars collect the data on the basis of established timelines and coding standards.
Patient Selection From ACS NSQIP, patients undergoing pancreatic resection for cancer from January 1, 2006, to December 31, 2009, were identified on the basis of Current Procedure Terminology24 and International Classification of Disease, Ninth Revision25 codes. Patients from the NCDB were identified using International Classification of Disease—Oncology, Third Revision26 codes. Patients were included in this study if a pancreas resection was performed for pathologically determined stage I-III pancreatic adenocarcinoma based on American Joint Committee on Cancer pathologic staging criteria.27 Patients were excluded from this study if treated with preoperative chemo- or radiotherapy, if chemotherapy was not recommended/administered because it was contraindicated (eg, comorbid conditions, advanced age, progression of tumor before administration), if the patient died before planned or recommended chemotherapy or if the patient was diagnosed at the reporting facility but treated elsewhere.
Adjuvant Chemotherapy Use We assessed both the receipt of adjuvant chemotherapy and, among patients who were treated, the number of days from surgery to chemotherapy initiation. The NCDB captures chemotherapy use if it was delivered as part of first course therapy (ie, not for recurrent disease or the treatment of another cancer). Time to adjuvant therapy was measured in days from the date of surgery to chemotherapy initiation. Treatment delay was defined as 70 or more days, which was the approximate upper quartile number of days from surgery to chemotherapy initiation and was considered clinically appropriate. Varying the definition of treatment delay resulted in similar findings.
Complications Twelve 30-day complications were separately assessed. First, a serious complication composite variable was evaluated that included major surgical and medical adverse events [organ space/deep surgical site infection (SSI), wound dehiscence, sepsis/shock, stroke or coma, myocardial infarction or cardiac arrest, venous thromboembolism (VTE), pneumonia, prolonged ventilation, reintubation, urinary tract infection, or renal failure). We also separately focused on the following individual complications: superficial SSI, organ space/deep SSI, wound dehiscence, prolonged ventilation, reintubation, pneumonia, VTE, renal failure, urinary tract infection (UTI), sepsis/septic shock, and return to the operating room. Patients were precluded from being categorized as having the following complications if the condition was documented preoperatively: wound infection, pneumonia, ventilator dependence/reintubation, or renal failure. Specific complication definitions are described elsewhere.28
Statistical Analysis
Descriptive statistics were calculated using the χ 2 test for categorical variables and Mann-Whitney U test to compare medians. We first assessed complication rates overall and among patients who did or did not receive adjuvant therapy. Next, we evaluated the association between operative complications and use of adjuvant therapy. C 2013 Lippincott Williams & Wilkins
Effect of Complications on Adjuvant Therapy Use
Hierarchical logistic regression models were developed to evaluate the association of postoperative complications on (1) adjuvant therapy omission and (2) treatment delay (≥70 days from surgery to initiation of chemotherapy). These models accounted for the hierarchical nature of patients nested within hospitals by including the hospital as a random effect. Variables were included in all models if they had a univariate association with the dependent variable at a P < 0.2 level or were considered clinically important. Final models included the following explanatory variables: age, sex, race, American Society of Anesthesiology (ASA) class, functional status, preoperative dyspnea, previous neurologic, cardiac or vascular event, body mass index, diabetes, chronic obstructive pulmonary disease, income status, AJCC (American Joint Committee on Cancer) stage, margin status, tumor grade, procedure and vascular reconstruction, and surgical volume status (Table 1). For both dependent variables assessed, postoperative complications were separately evaluated. Year of surgery was not associated with complications or adjuvant therapy administration and therefore was not included in the final analysis. As a sensitivity analysis, we further evaluated the association of serious morbidity on adjuvant therapy in a younger (
Postoperative Complications Reduce Adjuvant Chemotherapy Use in Resectable Pancreatic Cancer Ryan P. Merkow, MD, MS,∗ †‡ Karl Y. Bilimoria, MD, MS,∗ † James S. Tomlinson, MD, PhD,§ Jennifer L. Paruch, MD,∗ ‡ Jason B. Fleming, MD,¶ Mark S. Talamonti, MD,‡|| Clifford Y. Ko, MD, MS, MSHS,∗ § and David J. Bentrem, MD, MS†∗∗ Objective: To assess the impact of postoperative complications on the receipt of adjuvant chemotherapy. Background: Randomized trials have demonstrated that adjuvant chemotherapy is associated with improved long-term survival. However, pancreatic surgery is associated with significant morbidity and the degree to which complications limit subsequent treatment options is unknown. Methods: Patients from the American College of Surgeons National Surgical Quality Improvement Program and the National Cancer Data Base who underwent pancreatic resection for cancer were linked (2006–2009). The associations between complications and adjuvant chemotherapy use or treatment delay (≥70 days from surgery) were assessed using multivariable regression methods. Results: From 149 hospitals, 2047 patients underwent resection for stage I-III pancreatic adenocarcinoma of which 23.2% had at least 1 serious complication. Overall adjuvant chemotherapy receipt was 57.7%: 61.8% among patients not experiencing any complication and 43.6% among those who had a serious complication. Serious complications increased the likelihood of not receiving adjuvant therapy over twofold [odds ratio (OR) = 2.20, 95% confidence interval (CI): 1.73–2.80]. Specific complications associated with adjuvant chemotherapy omission were reintubation (OR = 7.79, 95% CI: 3.59–16.87), prolonged ventilation (OR = 5.92, 95% CI: 3.23–10.86), pneumonia (OR = 2.83, 95% CI: 1.63–4.90), sepsis/shock (OR = 2.76, 95% CI: 2.02–3.76), organ space/deep surgical site infection (OR = 2.19, 95% CI: 1.53–3.13), venous thromboembolism (OR = 1.92, 95% CI: 1.08–3.43), and urinary tract infection (OR = 1.61, 95% CI: 1.02–2.54). Serious complications also doubled the likelihood of delaying adjuvant treatment administration (OR = 2.08, 95% CI: 1.42–3.05). Sensitivity analysis in a younger, healthier patient cohort demonstrated similar associations. Conclusions: Postoperative complications are common following pancreatic surgery and are associated with adjuvant chemotherapy omission and treatment delays.
Keywords: ACS NSQIP, adjuvant therapy, complications, NCDB, pancreatic cancer, surgery (Ann Surg 2014;260:372–377)
I
n 2013, approximately 45 220 new cases of pancreatic cancer will be diagnosed and 38 460 people will succumb to the disease.1 By 2020, it is estimated that pancreatic cancer will become the second most common cause of cancer death.2 Although mortality after surgery has decreased markedly in recent years, complications remain common. More than 30% of patients experience at least 1 complication, and serious complications occur in up to 20% of patients.3,4 Although surgical resection is the primary component of multimodality therapy, pancreatic cancer is a biologically aggressive disease. To date, randomized trial data has demonstrated that adjuvant chemotherapy is associated with improved long-term survival when combined with surgical resection.5–8 Currently, national guidelines recommend adjuvant chemotherapy administration in the postoperative setting.9 Nevertheless, evidence suggests that more than 30% of patients who are eligible for systemic chemotherapy are never treated.10 Given pancreatic resection is associated with significant morbidity, the degree to which complications limit subsequent treatment options is unknown. If a strong relationship exists between complications and adjuvant therapy omission, theoretically, administering chemotherapy upfront before surgery could potentially increase the number of patients who would ultimately benefit from its effects.11,12 Our objectives were to determine the association of postoperative complications with adjuvant therapy receipt and, among patients who receive adjuvant therapy, assess the impact of complications on time to treatment.
METHODS Data Sources From the ∗ Division of Research and Optimal Patient Care, American College of Surgeons, Chicago, IL; †Surgical Outcomes and Quality Improvement Center and the Northwestern Institute for Comparative Effectiveness Research (NICER) in Oncology, Department of Surgery, Northwestern University Feinberg School of Medicine, Chicago, IL; ‡Department of Surgery, University of Chicago Pritzker School of Medicine, Chicago, IL; §Department of Surgery, University of California, Los Angeles and VA Greater Los Angeles Healthcare System, Los Angeles, CA; ¶Department of Surgical Oncology, MD Anderson Cancer Center, Houston, TX; ||Department of Surgery, Northshore University Health System, Evanston, IL; and ∗∗ Department of Surgery, Jesse Brown Veteran Affairs Medical Center, Chicago, IL. Disclosure: The authors have no conflicts of interest to disclose. This study was supported by a grant from the American Cancer Society (No. 280521) to R.P.M. and K.Y.B. J.L.P. is supported by a grant from Genetech. Presented at the 66th Society of Surgical Oncology Annual Cancer Symposium, March 8, 2013, National Harbor, MD. Reprints: Ryan P. Merkow, MD, MS, Division of Research and Optimal Patient Care, American College of Surgeons, 633 N Saint Clair Street, 22nd Floor, Chicago, IL 60611. E-mail: [email protected]. C 2013 by Lippincott Williams & Wilkins Copyright ISSN: 0003-4932/13/26002-0372 DOI: 10.1097/SLA.0000000000000378
372 | www.annalsofsurgery.com
The American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) and the National Cancer Data Base (NCDB) were linked to evaluate the effects of postoperative complications on adjuvant therapy use after pancreatic cancer surgery. ACS NSQIP contains detailed patient, demographic, comorbidity and complication data,13,14 while the NCDB collects cancer-specific tumor and treatment information.15 Because a unique patient identifier common to each data set does not exist, patients from ACS NSQIP were linked to the NCDB on the basis of a probabilistic matching algorithm using the following variables: American Hospital Association number, patient age, sex, and date of operation. The final match rate was 71%, which is comparable to the linkage of NSQIP with other data sources without the use of direct patient identifiers.16 The developmental history and current details of ACS NSQIP, including sampling strategy, data abstraction procedures, variables collected, complications assessed, and structure are described elsewhere.17–20 In brief, hospitals collect standardized and audited data by trained surgical clinical reviewers using standard ACS NSQIP Annals of Surgery r Volume 260, Number 2, August 2014
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Annals of Surgery r Volume 260, Number 2, August 2014
tools and definitions.21 Patients are followed for postoperative complications for 30 days after the index operation irrespective of whether the patient is an inpatient, has been discharged to his/her home or another facility, or has been readmitted to another hospital. The NCDB is a joint program of the American College of Surgeons, Commission on Cancer, and the American Cancer Society.22 NCDB is the largest cancer registry in the world, capturing more than 70% of all newly diagnosed malignancies in the United States from more than 1500 hospitals.23 Trained and audited cancer registrars collect the data on the basis of established timelines and coding standards.
Patient Selection From ACS NSQIP, patients undergoing pancreatic resection for cancer from January 1, 2006, to December 31, 2009, were identified on the basis of Current Procedure Terminology24 and International Classification of Disease, Ninth Revision25 codes. Patients from the NCDB were identified using International Classification of Disease—Oncology, Third Revision26 codes. Patients were included in this study if a pancreas resection was performed for pathologically determined stage I-III pancreatic adenocarcinoma based on American Joint Committee on Cancer pathologic staging criteria.27 Patients were excluded from this study if treated with preoperative chemo- or radiotherapy, if chemotherapy was not recommended/administered because it was contraindicated (eg, comorbid conditions, advanced age, progression of tumor before administration), if the patient died before planned or recommended chemotherapy or if the patient was diagnosed at the reporting facility but treated elsewhere.
Adjuvant Chemotherapy Use We assessed both the receipt of adjuvant chemotherapy and, among patients who were treated, the number of days from surgery to chemotherapy initiation. The NCDB captures chemotherapy use if it was delivered as part of first course therapy (ie, not for recurrent disease or the treatment of another cancer). Time to adjuvant therapy was measured in days from the date of surgery to chemotherapy initiation. Treatment delay was defined as 70 or more days, which was the approximate upper quartile number of days from surgery to chemotherapy initiation and was considered clinically appropriate. Varying the definition of treatment delay resulted in similar findings.
Complications Twelve 30-day complications were separately assessed. First, a serious complication composite variable was evaluated that included major surgical and medical adverse events [organ space/deep surgical site infection (SSI), wound dehiscence, sepsis/shock, stroke or coma, myocardial infarction or cardiac arrest, venous thromboembolism (VTE), pneumonia, prolonged ventilation, reintubation, urinary tract infection, or renal failure). We also separately focused on the following individual complications: superficial SSI, organ space/deep SSI, wound dehiscence, prolonged ventilation, reintubation, pneumonia, VTE, renal failure, urinary tract infection (UTI), sepsis/septic shock, and return to the operating room. Patients were precluded from being categorized as having the following complications if the condition was documented preoperatively: wound infection, pneumonia, ventilator dependence/reintubation, or renal failure. Specific complication definitions are described elsewhere.28
Statistical Analysis
Descriptive statistics were calculated using the χ 2 test for categorical variables and Mann-Whitney U test to compare medians. We first assessed complication rates overall and among patients who did or did not receive adjuvant therapy. Next, we evaluated the association between operative complications and use of adjuvant therapy. C 2013 Lippincott Williams & Wilkins
Effect of Complications on Adjuvant Therapy Use
Hierarchical logistic regression models were developed to evaluate the association of postoperative complications on (1) adjuvant therapy omission and (2) treatment delay (≥70 days from surgery to initiation of chemotherapy). These models accounted for the hierarchical nature of patients nested within hospitals by including the hospital as a random effect. Variables were included in all models if they had a univariate association with the dependent variable at a P < 0.2 level or were considered clinically important. Final models included the following explanatory variables: age, sex, race, American Society of Anesthesiology (ASA) class, functional status, preoperative dyspnea, previous neurologic, cardiac or vascular event, body mass index, diabetes, chronic obstructive pulmonary disease, income status, AJCC (American Joint Committee on Cancer) stage, margin status, tumor grade, procedure and vascular reconstruction, and surgical volume status (Table 1). For both dependent variables assessed, postoperative complications were separately evaluated. Year of surgery was not associated with complications or adjuvant therapy administration and therefore was not included in the final analysis. As a sensitivity analysis, we further evaluated the association of serious morbidity on adjuvant therapy in a younger (
