- Email: [email protected]
1022 Surgeon Characteristics Supersede Hospital Characteristics in Mortality After Urgent Colectomy
1021 Thirty-Day Mortality Underestimates Incidence of Death After Colorectal Surgery for Cancer Mohamed A. Adam, Zhifei Sun, Jina Kim, John Migaly, Christopher R. Mantyh Background: Thirty-day mortality after surgery is a critical quality metric of surgical care in the United States on which hospitals and surgeons are benchmarked. However, certain cancer surgical procedures are complex, with increased risk of complications after 30 days. Our aims were to compare incidence of 30-day vs. 90-day mortality after colorectal cancer surgery, and examine whether hospital performance ranking changes based on 30-day vs. 90-day mortality. Methods: Adult patients undergoing major surgery for non-metastatic colorectal cancers were included from the National Cancer Data Base (2004-2012). Mortality rates were compared within 30 vs. 90 days from surgery. Risk-adjusted hierarchical logistic regression models were employed to evaluate hospital performance based on 30-day and 90-day mortality. Hospitals were ranked into top (10%), middle (80%), and lowest (10%) performance groups based on each metric. Results: Among 186,763 patients included, 106,333 (57%) had colon cancer, 52,899 (28%) rectal cancer, and 27,531 (15%) rectosigmoid cancer. Composite 90-day mortality was nearly double 30-day mortality (4.3% vs. 2.5%, p<0.01); 90-day mortality was similarly higher across all cancer sites (Figure). With adjustment, older patient age, male gender, lower income, lack of insurance, comorbidities, advanced tumor stage, and treatment at non-academic centers were associated with mortality within 31-90 days. Based on risk-adjusted 30-day mortality, 116 (10%) hospitals ranked in the top performance group, 953 (80%) in the middle performance group, and 116 (10%) hospitals in lowest performance group. When considering risk-adjusted 90-day mortality, 180 (15%) hospitals changed performance ranking; among this group, 41% of hospitals in the top 30-day mortality performance group changed ranking to the middle performance group based on 90-day mortality. 37% of hospitals in the lowest 30-day mortality group changed ranking to the middle 90-day mortality performance group. 5% of hospitals in the middle 30-day performance group changed their ranking to the top 90-day performance group, with additional 5% changed ranking to the lowest 90-day performance group; Conclusions: A significant number of deaths after colorectal cancer surgery appear to occur beyond 30 days. Evaluation of hospital performance based on 30-day mortality is associated with misclassification of hospital ranking for 15% of hospitals. Evaluation of 90-day mortality may be a better quality metric in colorectal cancer surgery.
Table 1. Multivariate analysis of predictors of mortality after urgent colectomy.
1023
INTRODUCTION: Characterization of peritoneal and liver nodules is crucial to establish a precise evaluation of the cancer extension and to plan adequate therapy. Laparoscopic exploration can provide accurate assessment of the peritoneum cavity. Yet differentiating between cicatricial and carcinomatous lesions may be challenging. Probe based Confocal Laser Endomicroscopy (pCLE) is an emerging technique enabling real-time histology of living tissue. It could be used during laparoscopic procedures to help in the characterization of suspicious areas. This study aimed at evaluating the accuracy of CLE for the ex-vivo characterization of peritoneal carcinosis and liver samples. METHODS AND PROCEDURES: During this prospective, single-center trial, pCLE was performed ex vivo on fresh peritoneal and liver samples in 30 consecutive patients, after topical or IV administration of indocyanine green (ICG). The final diagnosis was obtained histologically, as per standard of care. pCLE image criteria for normal versus inflammatory/cicatricial versus malignant nodules were established from video sequences recorded on all samples by a pathologist expert in digestive disease and an expert in CLE image interpretation (phase I); these criteria were tested retrospectively on selected videos from these samples by two blind examiners: one surgeon and one pathoogist (phase II). The primary endpoints were values of accuracy in diagnosing malignant samples. RESULTS: Ex-vivo pCLE was successfully performed on 31 peritoneal nodules and 33 liver samples. 16 (53%) patients had had chemotherapy treatment prior to surgery. Among the 33 liver samples, 3 were excluded due to un-interpretable images. 19/ 30 (63%) were benign (healthy or cicatricial), 11/30 (37%) malignant. The sensitivity, specificity, positive predictive value, negative predictive value and accuracy to detect malignancy for the pathologist were 100%, 89%, 85%, 100% respectively. Among the 31 peritoneum specimens, 6/31 (19%) were benign, 12/31 (39%) inflammatory and 13/31 (42%) contained malignancy. Overall sensitivity, specificity, positive and negative predictive values to detect malignant nodules were 75%, 100%, 100% and 89%, respectively. Interobserver agreement was substantial (kappa 0.69). CONCLUSIONS: This preliminary study enabled to define interpretation criteria for peritoneum and liver and to validate them. CLE provides an important piece of additional information that correlate well with histopathology. Those results suggest that CLE is a very promising tool for the real-time characterization of peritoneum carcinomatosis and liver nodules.
1022 Surgeon Characteristics Supersede Hospital Characteristics in Mortality After Urgent Colectomy Richard S. Hoehn, Dennis J. Hanseman, Alex Chang, Meghan C. Daly, Audrey Ertel, Daniel E. Abbott, Shimul Shah, Ian M. Paquette OBJECTIVES: Urgent colectomy is a common procedure with a high mortality rate performed by a variety of surgeons and hospitals. We investigated whether surgeon or hospital characteristics more strongly predicted mortality after urgent colectomy. METHODS: The University HealthSystem Consortium was queried for adults undergoing urgent or emergent colectomy between 2009-2013 (n=18,213). Observed-to-expected (O/E) mortality ratios were created and surgeons and hospitals were grouped into quartiles accordingly. For hospital analysis, data were linked to the American Hospital Association (AHA) Annual Survey. Multiple logistic regression was used to determine patient and provider characteristics associated with in-hospital mortality. RESULTS: The overall mortality rate after urgent colectomy was 9.03%. Surgeons with the highest O/E mortality ratios tended to treat more patients who were black (Q4 vs Q1: 20% vs 15%), on Medicaid (17% vs 13%), and with extreme severity-of-illness (20% vs 12%; all p<0.001). These surgeons also performed more open cases (Q4 vs Q1: 87% vs 82%) and less total colectomies (9 vs 12%; all p<0.01). Similar associations were seen with hospital O/E ratios. Linkage with AHA data revealed that hospital characteristics did not differ significantly across O/E quartiles. Specifically, there were not significant correlations between hospital mortality rates and bed numbers, staffing (physicians, residents, nurses, technicians) or financial data (expenses, payroll, capital expenditures). On multivariate analysis (Table 1), age, male gender, Medicare insurance, and severity-of-illness were associated with postoperative mortality. Surgeon volume was inversely correlated with mortality while surgeons with a higher proportion of urgent cases (based on overall colectomy case load) had higher mortality. Hospital volume and case-mix were not statistically significant. CONCLUSION: Mortality is common after urgent colectomy and correlates most strongly with patient characteristics. Surgeon volume and practice patterns predicted differences in mortality whereas hospital factors did not. These data suggest that policies focusing solely on hospital volume may ignore other more important predictors of patient outcomes.
CLE images of healthy, inflammatory and malignant peritoneum (top line) and liver (bottom line) sample and corresponding histology. - ECM: Extra Cellular Matrix
S-1199
X : 10052$CH03 03-28-16 22:41:04 PDFd : 10052B : o
Page 1199
SSAT Abstracts
SSAT Abstracts
Accuracy of Confocal Laser Endomicroscopy for the Characterization of Peritoneal Carcinosis and Liver Nodules Brice Gayet, Angelo Pierangelo, David Fuks, Pierre Validire, Abdelali Benali
Table 1. Multivariate analysis of predictors of mortality after urgent colectomy.
1023
INTRODUCTION: Characterization of peritoneal and liver nodules is crucial to establish a precise evaluation of the cancer extension and to plan adequate therapy. Laparoscopic exploration can provide accurate assessment of the peritoneum cavity. Yet differentiating between cicatricial and carcinomatous lesions may be challenging. Probe based Confocal Laser Endomicroscopy (pCLE) is an emerging technique enabling real-time histology of living tissue. It could be used during laparoscopic procedures to help in the characterization of suspicious areas. This study aimed at evaluating the accuracy of CLE for the ex-vivo characterization of peritoneal carcinosis and liver samples. METHODS AND PROCEDURES: During this prospective, single-center trial, pCLE was performed ex vivo on fresh peritoneal and liver samples in 30 consecutive patients, after topical or IV administration of indocyanine green (ICG). The final diagnosis was obtained histologically, as per standard of care. pCLE image criteria for normal versus inflammatory/cicatricial versus malignant nodules were established from video sequences recorded on all samples by a pathologist expert in digestive disease and an expert in CLE image interpretation (phase I); these criteria were tested retrospectively on selected videos from these samples by two blind examiners: one surgeon and one pathoogist (phase II). The primary endpoints were values of accuracy in diagnosing malignant samples. RESULTS: Ex-vivo pCLE was successfully performed on 31 peritoneal nodules and 33 liver samples. 16 (53%) patients had had chemotherapy treatment prior to surgery. Among the 33 liver samples, 3 were excluded due to un-interpretable images. 19/ 30 (63%) were benign (healthy or cicatricial), 11/30 (37%) malignant. The sensitivity, specificity, positive predictive value, negative predictive value and accuracy to detect malignancy for the pathologist were 100%, 89%, 85%, 100% respectively. Among the 31 peritoneum specimens, 6/31 (19%) were benign, 12/31 (39%) inflammatory and 13/31 (42%) contained malignancy. Overall sensitivity, specificity, positive and negative predictive values to detect malignant nodules were 75%, 100%, 100% and 89%, respectively. Interobserver agreement was substantial (kappa 0.69). CONCLUSIONS: This preliminary study enabled to define interpretation criteria for peritoneum and liver and to validate them. CLE provides an important piece of additional information that correlate well with histopathology. Those results suggest that CLE is a very promising tool for the real-time characterization of peritoneum carcinomatosis and liver nodules.
1022 Surgeon Characteristics Supersede Hospital Characteristics in Mortality After Urgent Colectomy Richard S. Hoehn, Dennis J. Hanseman, Alex Chang, Meghan C. Daly, Audrey Ertel, Daniel E. Abbott, Shimul Shah, Ian M. Paquette OBJECTIVES: Urgent colectomy is a common procedure with a high mortality rate performed by a variety of surgeons and hospitals. We investigated whether surgeon or hospital characteristics more strongly predicted mortality after urgent colectomy. METHODS: The University HealthSystem Consortium was queried for adults undergoing urgent or emergent colectomy between 2009-2013 (n=18,213). Observed-to-expected (O/E) mortality ratios were created and surgeons and hospitals were grouped into quartiles accordingly. For hospital analysis, data were linked to the American Hospital Association (AHA) Annual Survey. Multiple logistic regression was used to determine patient and provider characteristics associated with in-hospital mortality. RESULTS: The overall mortality rate after urgent colectomy was 9.03%. Surgeons with the highest O/E mortality ratios tended to treat more patients who were black (Q4 vs Q1: 20% vs 15%), on Medicaid (17% vs 13%), and with extreme severity-of-illness (20% vs 12%; all p<0.001). These surgeons also performed more open cases (Q4 vs Q1: 87% vs 82%) and less total colectomies (9 vs 12%; all p<0.01). Similar associations were seen with hospital O/E ratios. Linkage with AHA data revealed that hospital characteristics did not differ significantly across O/E quartiles. Specifically, there were not significant correlations between hospital mortality rates and bed numbers, staffing (physicians, residents, nurses, technicians) or financial data (expenses, payroll, capital expenditures). On multivariate analysis (Table 1), age, male gender, Medicare insurance, and severity-of-illness were associated with postoperative mortality. Surgeon volume was inversely correlated with mortality while surgeons with a higher proportion of urgent cases (based on overall colectomy case load) had higher mortality. Hospital volume and case-mix were not statistically significant. CONCLUSION: Mortality is common after urgent colectomy and correlates most strongly with patient characteristics. Surgeon volume and practice patterns predicted differences in mortality whereas hospital factors did not. These data suggest that policies focusing solely on hospital volume may ignore other more important predictors of patient outcomes.
CLE images of healthy, inflammatory and malignant peritoneum (top line) and liver (bottom line) sample and corresponding histology. - ECM: Extra Cellular Matrix
S-1199
X : 10052$CH03 03-28-16 22:41:04 PDFd : 10052B : o
Page 1199
SSAT Abstracts
SSAT Abstracts
Accuracy of Confocal Laser Endomicroscopy for the Characterization of Peritoneal Carcinosis and Liver Nodules Brice Gayet, Angelo Pierangelo, David Fuks, Pierre Validire, Abdelali Benali