- Email: [email protected]
Hospital-free days, a novel composite endpoint, in patients undergoing pancreatic surgery
Mini Oral Session Abstracts
MO 87 MEASURED VS ESTIMATED BLOOD LOSS DURING PANCREATICODUODENECTOMY AND OTHER MAJOR ABDOMINAL OPERATIONS: INTERIM ANALYSIS L. Ghee, S. Thomas, G. Kowdley, S. Patel and S. Cunningham Saint Agnes Hospital and Cancer Institute, Ellicott City, MD, USA Objective: Estimated blood loss (EBL) is an important factor predicting clinical outcomes, but is frequently underand over-estimated, which can be dangerous for individual patients, and confounding for scoring systems, relying on EBL. Methods: We performed direct measurement of hemoglobin (hgb) levels of suction-canister volumes after collecting all blood from sponges and the field with dilute heparin-saline. Hgb levels were then used to calculate the measured blood loss (MBL), which was compared to the EBL, as estimated both by surgeons (sEBL) and anesthesiologists (aEBL). Power calculation predicted 83% power to detect a difference of 100 mL with a sample size of 35. An interim analysis was performed midway through the study. A paired t-test was used to compare MBL with EBL. Results: Of 23 eligible cases at interim analysis, pancreaticoduodenectomy (n = 8) was the most common. Median ASA score was 3 (range 2e4) and 96% of patients had comorbidities (median 3/patient). Median length of stay was 8 days (range 2e34), operative time was 5:14 (range 2:05e9:01), and complications occurred in 48%, and were Clavien grade >2 in 22%. The aEBL overestimated MBL by 192 mL (143%) on average, and was significantly greater than MBL (P = 0.004), while the sEBL was significantly less than MBL (P = 0.009). Conclusion: Surgeons underestimate and anesthesiologists overestimate EBL. This difference shown here is clinically substantial and statistically significant, and impacts not only immediate patient care but also the interpretation of scoring systems relying on EBL as a variable that may in fact be frequently inaccurate.
MO 88 DO DRAINS CONTRIBUTE TO PANCREATIC FISTULA? ANALYSIS OF OVER 5,000 PANCREATECTOMY PATIENTS IN A SINGLE YEAR R. Elkhoury, C. Kabir, V. Maker, M. Banulescu, M. Wasserman and A. Maker University of Illinois, Chicago, IL, USA Objective: Conflicting evidence exists from randomized controlled trials supporting both increased complications/ fistulae and improved outcomes with drain placement after pancreatectomy. The objective was to determine drain practice patterns in the U.S.A., and to identify if drain placement was a risk for fistula formation. Methods: Demographic, perioperative, and outcomes data were captured from the NSQIP 2014 database, including components of the fistula risk score. Fistulas were
HPB 2017, 19 (S1), S40eS108
S79
classified based on International Study Group definition. P < 0.05 was used for statistical significance in univariate analysis and entry criteria to adjusted logistic regression models. Results: Of 5013 pancreatectomy patients, 4343 (87%) underwent drain placement. When controlled for other factors, drain placement was associated with ducts <3 mm, soft glands, and blood transfusion within 72 h of surgery. Age, obesity, neoadjuvant radiation, INR, and malignant histology lost significance in the adjusted model. Drained patients experienced higher readmission rates (17 vs. 14%, p < 0.05); and experienced increased (20 vs. 8%, p < 0.01) and type A/B/C fistulae. Fistula was associated with obesity, no neoadjuvant chemotherapy, drain placement, <3 mm duct, soft gland, and longer operative times. Drain placement remained independently associated with fistula after both distal pancreatectomy (OR = 2.84 [1.70e4.75], p < 0.01) and pancreaticoduodenectomy (OR = 2.29 [1.28e4.11], p < 0.01). Conclusion: Drains are placed in the vast majority (87%) of pancreatectomy patients from >100 institutions; particularly those with soft glands, small ducts, and associated blood transfusions. When these factors are controlled for, drain placement is independently associated with clinically relevant fistulae in both distal and proximal pancreatectomy, raising questions regarding the utility of drain placement.
MO 89 HOSPITAL-FREE DAYS, A NOVEL COMPOSITE ENDPOINT, IN PATIENTS UNDERGOING PANCREATIC SURGERY A. Maiga, J. Wright, G. Edwards and K. Idrees Vanderbilt University Medical Center, Nashville, TN, USA Objective: Traditional post-operative metrics, i.e., length of stay (LOS) and readmission rates, individually do not fully quantify the totality of deviation from normalcy for patients. In this study, we utilize a novel composite endpoint, termed Hospital-Free Days (HFD), which incorporates post-operative outcomes into number of days patients spend outside of any healthcare facility after pancreatic resection (PR). Methods: We retrospectively reviewed 409 PR patients. Patient demographics, ASA class, Elixhauser Comorbidity Index (ECI), Surgical Apgar Score (SAS), and post-operative major complications (PMC) during index hospitalization were evaluated. HFD within 90 days from surgery were calculated by subtracting hospital LOS, readmission days, and days spent in rehabilitation/nursing facilities. Multivariable analysis (MVA) was used to examine association with HFD. Results: The median HFD after PR was 82 days (see Table). Patients with no PMC had a median HFD of 83 days compared to 77 days with a single PMC and 71 days for those with multiple PMC (p < 0.01). By MVA, age (p < 0.01), race (p < 0.01), ECI (p = 0.02), lower SAS (p < 0.01), and PMC (p < 0.01) were predictive of lower HFD. In patients without PMC during index hospitalization, lower SAS (p = 0.01) was the only significant factor associated with HFD. Conclusion: HFD is a novel, patient-centric, composite metric to quantify the true amount of time patients spend
S80
Mini Oral Session Abstracts
away from home in healthcare facilities after surgery. Patients who experience complications after PR have fewer HFD. HFD is an intuitive, single, clinically meaningful endpoint that enables providers to better establish expectations for prospective patients and serves as a surrogate measure of healthcare resource utilization.
Median 90-day HFDs (IQR)
p-Value
All patients
82 (75–84)
No complications
83 (78–84)
Ref
One complication
77 (68–80)
<0.01
Two or more complications
71 (65–75)
<0.01
MO 90 IMPACT OF PERIOPERATIVE CHANGES IN CA19-9 LEVELS IN PATIENTS WITH RESECTABLE AND BORDERLINE RESECTABLE PANCREATIC CANCER D. Wittmann, M. Aldakkak, E. P. Rajamanickam, K. K. Christians, M. Aburajab, B. George, P. S. Ritch, W. A. Hall, B. A. Erickson, D. B. Evans and S. Tsai Medical College of Wisconsin, Milwaukee, WI, USA Objective: Changes in CA19-9 values during treatment may have important prognostic implications. We examined the impact of perioperative changes in CA19-9 values in patients with pancreatic cancer (PC). Methods: CA19-9 values were classified as normal or elevated based on a cutoff of 35 U/mL. Patients with localized PC who received neoadjuvant therapy were grouped by the change in CA19-9 status from preop to postop: normal/normal, normal/elevated, elevated/normal, or elevated/elevated. Results: Of the 205 patients, 89 (44%) were normal/ normal, 4 (2%) were normal/elevated, 58 (28%) were elevated/normal, and 54 (26%) were elevated/elevated. Median postop CA19-9 for normal/normal, normal/ elevated, elevated/normal, elevated/elevated groups were 10, 68, 19, and 67, respectively. Median overall survival (OS) was 39 months; 48, not reached, 43, and 20 months in the normal/normal, normal/elevated, elevated/normal, and elevated/elevated groups, respectively (p < 0.001). In an adjusted hazards model, patients with elevated/elevated CA19-9 had a 2.82-fold (95% CI: 1.68e4.73) increased risk of death as compared to patients with normal/normal CA19-9. Conclusion: Following neoadjuvant therapy and surgery, the postoperative CA19-9 value is highly prognostic in patients with localized PC. If the preop CA19-9 is normal, it will likely remain normal postop. In contrast, only 50% of patients with elevated preop CA19-9 will normalize after surgery valuable information for investigators exploring alternative treatment sequencing for PC.
MO 91 IMPACT OF COMPLICATIONS ON HOSPITAL-FREE DAYS AFTER HEPATIC SURGERY J. Wright, A. Maiga, G. Edwards and K. Idrees Vanderbilt University Medical Center, Nashville, TN, USA Objective: Traditional post-operative metrics, i.e., length of stay (LOS) and readmission rates, individually do not fully quantify the totality of deviation from normalcy for patients. In this study, we utilize a novel measure, termed Hospital-Free Days (HFD), which incorporates postoperative outcomes into number of days patients spend outside of any healthcare facility after hepatic resection (HR). Methods: 463 HR patients over a 7-year period were retrospectively reviewed. Patient demographics, ASA class, Elixhauser Comorbidity Index (ECI), Surgical Apgar Score (SAS), and post-operative major complications (PMC) during index hospitalization were evaluated. HFD in the 90 days from surgery were calculated by subtracting hospital LOS, readmission days, and days spent in rehabilitation/nursing facilities. Multivariable analysis (MVA) was utilized to examine association with HFD. Results: The median HFD after HR was 85 days (see table). Patients without PMC had a median HFD of 85 days compared to 78 days with a single PMC and 72 days for those with multiple PMC (p < 0.01). Age (p = 0.04), lower SAS (p < 0.01), and PMC (p < 0.01) were predictive of the lower HFD on MVA. In patients without PMC during the index hospitalization, SAS (p = 0.0007) and ECI (p = 0.02) were predictive of lower HFD. Conclusion: HFD is an intuitive, singular, and clinically meaningful endpoint to quantify the true amount of time a patient spends away from home in healthcare facilities. Complications have a significant impact on lowering HFD following HR. HFD is a novel, patient-centric metric which allows providers to better establish expectations for prospective patients and serves as a surrogate measure of healthcare resource utilization.
HPB 2017, 19 (S1), S40eS108
MO 87 MEASURED VS ESTIMATED BLOOD LOSS DURING PANCREATICODUODENECTOMY AND OTHER MAJOR ABDOMINAL OPERATIONS: INTERIM ANALYSIS L. Ghee, S. Thomas, G. Kowdley, S. Patel and S. Cunningham Saint Agnes Hospital and Cancer Institute, Ellicott City, MD, USA Objective: Estimated blood loss (EBL) is an important factor predicting clinical outcomes, but is frequently underand over-estimated, which can be dangerous for individual patients, and confounding for scoring systems, relying on EBL. Methods: We performed direct measurement of hemoglobin (hgb) levels of suction-canister volumes after collecting all blood from sponges and the field with dilute heparin-saline. Hgb levels were then used to calculate the measured blood loss (MBL), which was compared to the EBL, as estimated both by surgeons (sEBL) and anesthesiologists (aEBL). Power calculation predicted 83% power to detect a difference of 100 mL with a sample size of 35. An interim analysis was performed midway through the study. A paired t-test was used to compare MBL with EBL. Results: Of 23 eligible cases at interim analysis, pancreaticoduodenectomy (n = 8) was the most common. Median ASA score was 3 (range 2e4) and 96% of patients had comorbidities (median 3/patient). Median length of stay was 8 days (range 2e34), operative time was 5:14 (range 2:05e9:01), and complications occurred in 48%, and were Clavien grade >2 in 22%. The aEBL overestimated MBL by 192 mL (143%) on average, and was significantly greater than MBL (P = 0.004), while the sEBL was significantly less than MBL (P = 0.009). Conclusion: Surgeons underestimate and anesthesiologists overestimate EBL. This difference shown here is clinically substantial and statistically significant, and impacts not only immediate patient care but also the interpretation of scoring systems relying on EBL as a variable that may in fact be frequently inaccurate.
MO 88 DO DRAINS CONTRIBUTE TO PANCREATIC FISTULA? ANALYSIS OF OVER 5,000 PANCREATECTOMY PATIENTS IN A SINGLE YEAR R. Elkhoury, C. Kabir, V. Maker, M. Banulescu, M. Wasserman and A. Maker University of Illinois, Chicago, IL, USA Objective: Conflicting evidence exists from randomized controlled trials supporting both increased complications/ fistulae and improved outcomes with drain placement after pancreatectomy. The objective was to determine drain practice patterns in the U.S.A., and to identify if drain placement was a risk for fistula formation. Methods: Demographic, perioperative, and outcomes data were captured from the NSQIP 2014 database, including components of the fistula risk score. Fistulas were
HPB 2017, 19 (S1), S40eS108
S79
classified based on International Study Group definition. P < 0.05 was used for statistical significance in univariate analysis and entry criteria to adjusted logistic regression models. Results: Of 5013 pancreatectomy patients, 4343 (87%) underwent drain placement. When controlled for other factors, drain placement was associated with ducts <3 mm, soft glands, and blood transfusion within 72 h of surgery. Age, obesity, neoadjuvant radiation, INR, and malignant histology lost significance in the adjusted model. Drained patients experienced higher readmission rates (17 vs. 14%, p < 0.05); and experienced increased (20 vs. 8%, p < 0.01) and type A/B/C fistulae. Fistula was associated with obesity, no neoadjuvant chemotherapy, drain placement, <3 mm duct, soft gland, and longer operative times. Drain placement remained independently associated with fistula after both distal pancreatectomy (OR = 2.84 [1.70e4.75], p < 0.01) and pancreaticoduodenectomy (OR = 2.29 [1.28e4.11], p < 0.01). Conclusion: Drains are placed in the vast majority (87%) of pancreatectomy patients from >100 institutions; particularly those with soft glands, small ducts, and associated blood transfusions. When these factors are controlled for, drain placement is independently associated with clinically relevant fistulae in both distal and proximal pancreatectomy, raising questions regarding the utility of drain placement.
MO 89 HOSPITAL-FREE DAYS, A NOVEL COMPOSITE ENDPOINT, IN PATIENTS UNDERGOING PANCREATIC SURGERY A. Maiga, J. Wright, G. Edwards and K. Idrees Vanderbilt University Medical Center, Nashville, TN, USA Objective: Traditional post-operative metrics, i.e., length of stay (LOS) and readmission rates, individually do not fully quantify the totality of deviation from normalcy for patients. In this study, we utilize a novel composite endpoint, termed Hospital-Free Days (HFD), which incorporates post-operative outcomes into number of days patients spend outside of any healthcare facility after pancreatic resection (PR). Methods: We retrospectively reviewed 409 PR patients. Patient demographics, ASA class, Elixhauser Comorbidity Index (ECI), Surgical Apgar Score (SAS), and post-operative major complications (PMC) during index hospitalization were evaluated. HFD within 90 days from surgery were calculated by subtracting hospital LOS, readmission days, and days spent in rehabilitation/nursing facilities. Multivariable analysis (MVA) was used to examine association with HFD. Results: The median HFD after PR was 82 days (see Table). Patients with no PMC had a median HFD of 83 days compared to 77 days with a single PMC and 71 days for those with multiple PMC (p < 0.01). By MVA, age (p < 0.01), race (p < 0.01), ECI (p = 0.02), lower SAS (p < 0.01), and PMC (p < 0.01) were predictive of lower HFD. In patients without PMC during index hospitalization, lower SAS (p = 0.01) was the only significant factor associated with HFD. Conclusion: HFD is a novel, patient-centric, composite metric to quantify the true amount of time patients spend
S80
Mini Oral Session Abstracts
away from home in healthcare facilities after surgery. Patients who experience complications after PR have fewer HFD. HFD is an intuitive, single, clinically meaningful endpoint that enables providers to better establish expectations for prospective patients and serves as a surrogate measure of healthcare resource utilization.
Median 90-day HFDs (IQR)
p-Value
All patients
82 (75–84)
No complications
83 (78–84)
Ref
One complication
77 (68–80)
<0.01
Two or more complications
71 (65–75)
<0.01
MO 90 IMPACT OF PERIOPERATIVE CHANGES IN CA19-9 LEVELS IN PATIENTS WITH RESECTABLE AND BORDERLINE RESECTABLE PANCREATIC CANCER D. Wittmann, M. Aldakkak, E. P. Rajamanickam, K. K. Christians, M. Aburajab, B. George, P. S. Ritch, W. A. Hall, B. A. Erickson, D. B. Evans and S. Tsai Medical College of Wisconsin, Milwaukee, WI, USA Objective: Changes in CA19-9 values during treatment may have important prognostic implications. We examined the impact of perioperative changes in CA19-9 values in patients with pancreatic cancer (PC). Methods: CA19-9 values were classified as normal or elevated based on a cutoff of 35 U/mL. Patients with localized PC who received neoadjuvant therapy were grouped by the change in CA19-9 status from preop to postop: normal/normal, normal/elevated, elevated/normal, or elevated/elevated. Results: Of the 205 patients, 89 (44%) were normal/ normal, 4 (2%) were normal/elevated, 58 (28%) were elevated/normal, and 54 (26%) were elevated/elevated. Median postop CA19-9 for normal/normal, normal/ elevated, elevated/normal, elevated/elevated groups were 10, 68, 19, and 67, respectively. Median overall survival (OS) was 39 months; 48, not reached, 43, and 20 months in the normal/normal, normal/elevated, elevated/normal, and elevated/elevated groups, respectively (p < 0.001). In an adjusted hazards model, patients with elevated/elevated CA19-9 had a 2.82-fold (95% CI: 1.68e4.73) increased risk of death as compared to patients with normal/normal CA19-9. Conclusion: Following neoadjuvant therapy and surgery, the postoperative CA19-9 value is highly prognostic in patients with localized PC. If the preop CA19-9 is normal, it will likely remain normal postop. In contrast, only 50% of patients with elevated preop CA19-9 will normalize after surgery valuable information for investigators exploring alternative treatment sequencing for PC.
MO 91 IMPACT OF COMPLICATIONS ON HOSPITAL-FREE DAYS AFTER HEPATIC SURGERY J. Wright, A. Maiga, G. Edwards and K. Idrees Vanderbilt University Medical Center, Nashville, TN, USA Objective: Traditional post-operative metrics, i.e., length of stay (LOS) and readmission rates, individually do not fully quantify the totality of deviation from normalcy for patients. In this study, we utilize a novel measure, termed Hospital-Free Days (HFD), which incorporates postoperative outcomes into number of days patients spend outside of any healthcare facility after hepatic resection (HR). Methods: 463 HR patients over a 7-year period were retrospectively reviewed. Patient demographics, ASA class, Elixhauser Comorbidity Index (ECI), Surgical Apgar Score (SAS), and post-operative major complications (PMC) during index hospitalization were evaluated. HFD in the 90 days from surgery were calculated by subtracting hospital LOS, readmission days, and days spent in rehabilitation/nursing facilities. Multivariable analysis (MVA) was utilized to examine association with HFD. Results: The median HFD after HR was 85 days (see table). Patients without PMC had a median HFD of 85 days compared to 78 days with a single PMC and 72 days for those with multiple PMC (p < 0.01). Age (p = 0.04), lower SAS (p < 0.01), and PMC (p < 0.01) were predictive of the lower HFD on MVA. In patients without PMC during the index hospitalization, SAS (p = 0.0007) and ECI (p = 0.02) were predictive of lower HFD. Conclusion: HFD is an intuitive, singular, and clinically meaningful endpoint to quantify the true amount of time a patient spends away from home in healthcare facilities. Complications have a significant impact on lowering HFD following HR. HFD is a novel, patient-centric metric which allows providers to better establish expectations for prospective patients and serves as a surrogate measure of healthcare resource utilization.
HPB 2017, 19 (S1), S40eS108