Artificial Neural Network (ANN) Is Superior to the Pre-Endoscopic Rockall Score in Predicting Which Patients Will Benefit From Urgent EGD in Acute Upper GI Bleeding (UGIB)

Artificial Neural Network (ANN) Is Superior to the Pre-Endoscopic Rockall Score in Predicting Which Patients Will Benefit From Urgent EGD in Acute Upper GI Bleeding (UGIB)

Abstracts 912.04 Artificial Neural Network (ANN) Is Superior to the PreEndoscopic Rockall Score in Predicting Which Patients Will Benefit From Urgent...

95KB Sizes 1 Downloads 39 Views

Abstracts

912.04 Artificial Neural Network (ANN) Is Superior to the PreEndoscopic Rockall Score in Predicting Which Patients Will Benefit From Urgent EGD in Acute Upper GI Bleeding (UGIB) Ananya Das, Farees Farooq, Richard Wong, Tamir Ben-Menachem, Gregory Cooper, Amitabh Chak, Michael Sivak Introduction: In practice, not all patients with acute UGIB can receive urgent EGD (within 12 hours). A model for predicting need for urgent EGD based on nonendoscopic, clinical variables would be an important tool in patient triage. We have previously shown that ANN accurately predicts outcome in acute LGI bleeding (Lancet 2003;362:1261). Methods: We constructed an ANN to predict endoscopic stigmata of recent hemorrhage (SRH) and need for endoscopic therapy (ET) using prospectively collected data from 194 patients at an academic medical center who presented with acute UGIB over a 6-month period (training group). Confirmed variceal bleeding was excluded. The ANN model was a multi-layered perceptron network trained by back propagation using pre-endoscopic clinical input variables available at the time of triage. The trained ANN was applied to an internal validation (IV) group of 193 patients with UGIB during the same time period. The ANN was then applied to an external validation (EV) group of 200 patients at a tertiary care center in a different city over a 6-month period. The performance of the ANN was also compared to the widely used Rockall score. A pre-endoscopic Rockall score of 0 is considered low risk for adverse outcome. Results: The clinical characteristics of the EV group were different then the training and IV groups (increased men, CAD, cirrhosis, hematemesis, shock, ICU admit, rebleeding, and death). Using ANN, the areas under the ROC curve for predicting SRH and ET were 0.94 & 0.84 in the IV group and 0.81 & 0.78 in the EV group, respectively. ANN had better specificity and predictive values than pre-endoscopic Rockall score in both validation groups and was comparable to total (post-endoscopic) Rockall score. Conclusions: ANN is an effective tool in the pre-endoscopic triage of patients with acute UGIB, even in a dissimilar external cohort. ANN is more specific than the preendoscopic Rockall score in identifying patients who may benefit from urgent EGD and may be used to exclude low risk patients.

Health Planning and Development. We used these data to identify all patients hospitalized in California with gastric or duodenal ulcer bleeding in 1995. 1,070 of total 15,503 such patients had prior admissions for PUD bleeding between 1991-4. Therefore 14,473 patients were admitted with an initial PUD bleed in 1995. We compared 367 (2.5%) patients who were readmitted with rebleeding during 19952000 to 14,106 patients who did not rebleed using multivariate analysis which adjusted for age, gender, length of stay, level of care, disposition, disease severity, and time to endoscopic therapy. We extrapolated income from the ZIP code of admission using data from the 2000 US Census. We used chi-square and student t-tests for univariate analysis and logistic regression for multivariate analysis with alpha of 0.05 considered significant. This study was approved by the Statewide and the University IRB. Results: Patients readmitted with rebleeding had significantly shorter (3.8 G 2.7d vs. 4.8 G 6.6d, p!0.0001) and less expensive ($10,615 G 10,625 vs. $13,758 G 25,455, p!0.0001) initial hospital stay, lower annual household income ($44,593 G 15,726 vs. $48,163 G 18,657, p!0.0001) compared to non rebleeders. The rebleed rates also differed significantly according to the funding source, p Z 0.0024. Multivariate analysis showed Native American/Eskimo race (OR Z 4.0, 95% CI: 1.7-9.4, p Z 0.0015) and black race (OR Z 1.5, 95% CI: 1.2-2.0, p Z 0.0005) to have significantly higher risk of rebleeding compared to Caucasians. The same analysis showed HMO funding (OR Z 0.76, 95% CI: 0.62-0.95, p Z 0.014) and higher income (OR Z 0.94, 95% CI: 0.90-0.98, p Z 0.0078) to be associated with lower risk of rebleeding. Conclusion: Only 2.5% of patients admitted with PUD bleeding were readmitted for recurrent PUD bleeding during the 5 year follow up. Risk factors for rebleeding were being black, Native American or Eskimo race, while HMO funding and higher income were associated with lower risk of rebleeding.

High Pre-endoscopic Rockall (> 0) vs. ANN in predicting SRH

912.06 Changing Epidemiology of Upper Gastrointestinal Bleeding (UGI) in Patients Hospitalized in California: Population Based Study From 1991-2000 John Lee, Sanjay Reddy, Namgyal Kyulo High Pre-endoscopic Rockall (> 0) vs. ANN in predicting need for ET

Background: We report the changing incidence and mortality of UGI bleeding in California. Methods: California law mandates that all hospitals report details of hospital discharges to the Office of Statewide Health Planning and Development. We used these data to identify all patients hospitalized with UGI bleeding (ICD codes 4560-2, 5302, 5310-49) from 1991-2000. Subjects were categorized as variceal (n Z 5,979) or non variceal (n Z 193,194) bleeders by discharge diagnosis. Rebleeding was defined as a repeat admission within one year for UGI bleeding occurring from the same source as the initial bleed. Age and gender adjusted incidence rates were calculated using data from the 2000 US Census. This study was approved by the Statewide and the University IRB. Results: Rebleeding decreased significantly for variceal bleeders from 10.42% in 1991 to 5.51% in 2000, p!0.05. In contrast, non variceal bleeders had an initial increase in rebleeding from 5.0% in 1991 to 7.99% 1997, but decreased steadily afterwards to 6.10% in 2000. The hospital mortality rate decreased significantly for non variceal bleeders from 9.46% in 1991 to 4.84% in 2000, p!0.05, but not for non variceal bleeders (4.02% in 1991 to 3.72% in 2000, p>0.05.) Conclusion: The incidences of hospitalizations for variceal and non variceal bleeding decreased steadily in California with the largest drop occurring in 1994-5. Rate of rebleeding decreased by half for varices, but increased for non variceal bleeding from 1991 to 1997, then fell afterwards coinciding with the introduction of proton pump inhibitor therapy. Hospital mortality decreased by half for variceal bleeding but did not change for non variceal bleeding.

912.05 Long Term Population Study of Prevalence and Risk Factors for Rebleeding in Patients with Peptic Ulcer Disease (PUD) John Lee, Sanjay Reddy, Namgyal Kyulo Background: Most data on rebleeding are reported from tertiary care centers over a short follow up period and may not reflect community trends. We sought to determine the prevalence and risk factors for being hospitalized with rebleeding over a 5 year period for patients in California. Methods: California Health Data and Advisory Council Consolidation Act mandates that all hospitals licensed in California report details of all hospitalizations to the California Office of Statewide

www.mosby.com/gie

Volume 61, No. 5 : 2005 GASTROINTESTINAL ENDOSCOPY AB87