Artificial intelligence-assisted system for the assessment of Forrest classification of peptic ulcer bleeding: a multicenter diagnostic study.

BACKGROUND: Inaccurate Forrest classification may significantly affect clinical outcomes, especially in high risk patients. Therefore, this study aimed to develop a real-time deep convolutional neural network (DCNN) system to assess the Forrest classification of peptic ulcer bleeding (PUB). METHODS: A training dataset (3868 endoscopic images) and an internal validation dataset (834 images) were retrospectively collected from the 900th Hospital, Fuzhou, China. In addition, 521 images collected from four other hospitals were used for external validation. Finally, 46 endoscopic videos were prospectively collected to assess the real-time diagnostic performance of the DCNN system, whose diagnostic performance was also prospectively compared with that of three senior and three junior endoscopists. RESULTS: The DCNN system had a satisfactory diagnostic performance in the assessment of Forrest classification, with an accuracy of 91.2% (95%CI 89.5%-92.6%) and a macro-average area under the receiver operating characteristic curve of 0.80 in the validation dataset. Moreover, the DCNN system could judge suspicious regions automatically using Forrest classification in real-time videos, with an accuracy of 92.0% (95%CI 80.8%-97.8%). The DCNN system showed more accurate and stable diagnostic performance than endoscopists in the prospective clinical comparison test. This system helped to slightly improve the diagnostic performance of senior endoscopists and considerably enhance that of junior endoscopists. CONCLUSION: The DCNN system for the assessment of the Forrest classification of PUB showed satisfactory diagnostic performance, which was slightly superior to that of senior endoscopists. It could therefore effectively assist junior endoscopists in making such diagnoses during gastroscopy.

Reassessment of the predictive value of the Forrest classification for peptic ulcer rebleeding and mortality: can classification be simplified?

BACKGROUND AND STUDY AIMS: This study aimed to reassess whether the Forrest classification is still useful for the prediction of rebleeding and mortality in peptic ulcer bleedings and, based on this, whether the classification could be simplified. PATIENTS AND METHODS: Prospective registry data on peptic ulcer bleedings were collected and categorized according to the Forrest classification. The primary outcomes were 30-day rebleeding and all-cause mortality rates. Receiver operating characteristic curves were used to test whether simplification of the Forrest classification into high risk (Forrest Ia), increased risk (Forrest Ib-IIc), and low risk (Forrest III) classes could be an alternative to the original classification. RESULTS: In total, 397 patients were included, with 18 bleedings (4.5%) being classified as Forrest Ia, 73 (18.4%) as Forrest Ib, 86 (21.7%) as Forrest IIa, 32 (8.1%) as Forrest IIb, 59 (14.9%) as Forrest IIc, and 129 (32.5%) as Forrest III. Rebleeding occurred in 74 patients (18.6%). Rebleeding rates were highest in Forrest Ia peptic ulcers (59%). The odds ratios for rebleeding among Forrest Ib-IIc ulcers were similar. In subgroup analysis, predicting rebleeding using the Forrest classification was more reliable for gastric ulcers than for duodenal ulcers. The simplified Forrest classification had similar test characteristics to the original Forrest classification. CONCLUSION: The Forrest classification still has predictive value for rebleeding of peptic ulcers, especially for gastric ulcers; however, it does not predict mortality. Based on these results, a simplified Forrest classification is proposed. However, further studies are needed to validate these findings.

Medical and endoscopic treatment in peptic ulcer bleeding: a national German survey.

OBJECTIVE: Peptic ulcers are the leading cause of upper gastrointestinal (GI) bleeding. The aim of this study was the evaluation of the recent clinical practice in drug therapy and endoscopic treatment of ulcer bleedings in Germany and to compare the results with the medical standard. METHODS: A structured questionnaire (cross-sectional study) was sent to 1371 German hospitals that provide an emergency service for upper GI bleeding. The project was designed similar to a nationwide inquiry in France in 2001. Forty-four questions concerning the following topics were asked: hospital organisation, organisation of emergency endoscopy service, endoscopic and drug therapy of ulcer bleeding, endoscopic treatment of variceal bleeding. Return of the questionnaires was closed in August 2004. RESULTS: Response rate was 675 / 1371 (49 %). Mean hospitals size was < 200 beds, 49 % (n = 325) had basic care level. 92 % provided a 24-hour endoscopy service, specialized nurses were available in 75 %. Fiberscopes were used only in 15 %. A mean of 10 +/- 12 (range: 0 - 160) bleeding cases/month were treated, 6 +/- 6 cases per month (60 %) were ulcer bleedings. Endoscopy was performed in 72 % immediately after stabilization but in all cases within 24 hours. The Forrest classification was used in 99 % whereas prognostic scores were applied only in 3 %. Forrest Ia,/Ib/IIa/IIb/IIc/III ulcers were indications for endoscopic therapy in 99 %/ 99 %/ 90 %/ 58 %/ 4 %/ 2 % respectively. Favoured initial treatment was injection (diluted epinephrine, mean volume 17 +/- 13 mL/lesion) followed by clipping. In re-bleedings, 93 % tried endoscopic treatment again. Scheduled re-endoscopy was performed in 63 %. PPI were used in 99.6 %, 85 % administered standard dose twice daily. PPI administration was changed from intravenous to oral with the end of fasting in nearly all hospitals. PPI administration schemes can be improved. Indications for Helicobacter pylori eradication followed rational principles. CONCLUSION: Medical and endoscopic treatment of bleeding ulcers reached a high standard, although some therapeutic strategies leave room for improvement. Bigger hospitals tend to be closer to the medical standard.

Moderate and high affinity serotonin reuptake inhibitors increase the risk of upper gastrointestinal toxicity.

OBJECTIVE: Serotonin release from platelets is important for regulating hemostasis. Some prior studies suggest an association between use of selective serotonin reuptake inhibitors and gastrointestinal bleeding and a possible synergistic effect of these medications with non-steroidal anti-inflammatory drugs (NSAIDs). This study examined the effect of medications that inhibit serotonin uptake on upper gastrointestinal toxicity. METHODS: 359 case subjects hospitalized for upper gastrointestinal bleeding, perforation, or benign gastric outlet obstruction were recruited from 28 hospitals. 1889 control subjects were recruited by random digit dialing from the same region. Data were collected during structured telephone interviews. Antidepressant medications were characterized according to their affinity for serotonin receptors. Exposure to medications required use on at least 1 day during the week prior to the index date. RESULTS: Any moderate or high affinity serotonin reuptake inhibitor (MHA-SRI) use was reported by 61 cases (17.1%) and 197 controls (10.4%). After adjusting for potential confounders, MHA-SRI use was associated with a significantly increased odds of hospitalization for upper gastrointestinal toxicity (adjusted OR = 2.0, 95%CI 1.4-3.0). A dose-response relationship in terms of affinity for serotonin uptake receptors was not observed (p = 0.17). No statistical interaction was observed for use of high dose NSAIDs or aspirin concomitantly with MHA-SRIs (p = 0.5). When MHA-SRIs were used concomitantly with high dose NSAIDs, the adjusted odds ratio for the association with upper gastrointestinal toxicity was 3.5 (95%CI 1.9-6.6). CONCLUSIONS: Use of MHA-SRIs is associated with an increased risk of hospitalization for upper gastrointestinal toxicity.

[The role of endoscopy in diagnosis and treatment of gastroduodenal bleedings].

The experience in diagnosis and treatment of 469 patients with gastroduodenal ulcer bleedings is presented. Sensitivity of endoscopy in detection of bleeding source was 98.8%, in detection of ulcer size -- 93.4%, specificity -- 80.6%. It is demonstrated that the most frequently recurrence of bleeding occurs when ulcer size is more than 1 cm. Location of ulcer on posterior, posterior-superior and posterior-inferior walls of the duodenal bulb is an unfavorable prognostic symptom (rebleeding occurs in 17.1, 30.0 and 12.5% cases, respectively). Types Forrest-Ia and Forrest-IIa are especially dangerous for recurrence (21.4 and 15.6%, respectively). In types Forrest-Ib and Forrest-IIb rebleeding occurs less frequently (4 and 6%, respectively). In types Forrest-Ia and Forrest-Ib rebleeding occurs mainly on the first day, rebleeding in types Forrest-IIa occurs more often on the first-second day, however it is possible up to day 7. Rebleeding in types Forrest-IIb occurs also mainly during the first two days. Endoscopic hemostasis (electrocoagulation and injection therapy) was performed in 40 patients. In 6 (15%) cases hemostasis was not achieved. In 8 (20%) cases endoscopic hemostasis led to a temporary effect, i.e. rebleeding was seen within 24-72 hours after endoscopic procedure. Stable hemostasis with endoscopy was achieved in 26 (65%) patients.

[Clinical usefulness of Rockall scoring system in patients with bleeding peptic ulcer--comparison with Forrest classification].

BACKGROUND/AIMS: The Rockall risk assessment score was developed to predict the risk of rebleeding and death in patients with upper GI hemorrhage. The validity of this score, however, was not established in Korea. We tried to assess the reliability of the Rockall score to predict outcomes in patients with bleeding peptic ulcer. METHODS: Medical records of 175 patients with benign peptic ulcer bleeding treated in Samsung Medical Center from January 2000 to May 2003 were retrospectively analyzed. They were classified into three groups: no rebleeding rebleeding, and death and mean Rockall score was compared. Forrest classification was also compared with the Rockall score regarding the clinical usefulness of predicting poor outcomes in patients with bleeding peptic ulcer. RESULTS: One hundred forty five patients did not show rebleeding, with mean Rockall score of 3.5 (SD=1.5). On the other hand, rebleeding occurred in 25 patients and the mean score was 6.4 (SD=1.44). There were 13 deaths with mean score of 7.0 (SD=1.08). The differences between the three groups were significant (p<0.001). In multivariate analysis, Rockall score was a independent risk factor of rebleeding and mortality (odds ratio, OR=2.73 and OR=8.74). CONCLUSIONS: The Rockall scoring system is useful to predict poor outcome such as rebleeding and death in patients with bleeding peptic ulcer.

Clinical Usefulness of Rockall Scoring System in Patients with Bleeding Peptic Ulcer-Comparison with Forrest Classification / 대한소화기학회지

BACKGROUND/AIMS: The Rockall risk assessment score was developed to predict the risk of rebleeding and death in patients with upper GI hemorrhage. The validity of this score, however, was not established in Korea. We tried to assess the reliability of the Rockall score to predict outcomes in patients with bleeding peptic ulcer. METHODS: Medical records of 175 patients with benign peptic ulcer bleeding treated in Samsung Medical Center from January 2000 to May 2003 were retrospectively analyzed. They were classified into three groups: no rebleeding rebleeding, and death and mean Rockall score was compared. Forrest classification was also compared with the Rockall score regarding the clinical usefulness of predicting poor outcomes in patients with bleeding peptic ulcer. RESUTLS: One hundred forty five patients did not show rebleeding, with mean Rockall score of 3.5 (SD=1.5). On the other hand, rebleeding occurred in 25 patients and the mean score was 6.4 (SD=1.44). There were 13 deaths with mean score of 7.0 (SD=1.08). The differences between the three groups were significant (p<0.001). In multivariate analysis, Rockall score was a independent risk factor of rebleeding and mortality (odds ratio, OR=2.73 and OR=8.74). CONCLUSIONS: The Rockall scoring system is useful to predict poor outcome such as rebleeding and death in patients with bleeding peptic ulcer.

The role of endoscopic Doppler US in patients with peptic ulcer bleeding.

BACKGROUND: Stigmata of recent hemorrhage are important prognostic signs for patients with ulcer bleeding, but these are subjective findings. This study evaluated the additional diagnostic value of Doppler US assessment in patients with a bleeding peptic ulcer. METHODS: A prospective, multicenter study was performed of patients with ulcer bleeding. Stigmata of recent hemorrhage were classified according to the Forrest classification, after which the ulcer was assessed by using an endoscopic Doppler US system. Patients with a Forrest Ib-IIb ulcer with a positive Doppler signal received endoscopic therapy. Patients with a Forrest IIc-III ulcer with a positive Doppler signal were allocated randomly to endoscopic therapy or no therapy. No ulcer without a Doppler signal was treated. RESULTS: A total of 80 patients were enrolled. Of the Forrest Ib-IIb ulcers, 82% had a positive Doppler signal. Of the Forrest IIc-III ulcers, 53% had a positive Doppler signal. There was no difference in the rates of recurrent bleeding, surgery, or mortality between the group with Forrest Ib-IIb ulcers and between the Forrest IIc-III group with and without Doppler signal, but there was little power in the sample size to detect differences. Bleeding recurred in 3 patients without a Doppler signal. Recurrent bleeding was more frequent in the group in which a Doppler signal was still present immediately after endoscopic therapy (3/11 vs. 1/27; p=0.06). CONCLUSIONS: This study did not substantiate a role for endoscopic Doppler US when this was added to the Forrest classification for making clinical decisions in patients with ulcer bleeding.

[Treatment of upper gastrointestinal bleeding caused by peptic ulcer]. / A peptikus fekélyból származó gastroduodenalis vérzések kezeléséról.

INTRODUCTION: The diagnosis and treatment of gastrointestinal bleeding represent a major problem even today. Nearly 50% of upper gastrointestinal bleedings are originating from peptic ulcers of different locations. Thanking to the modern gastroenterological treatment the number of elective operations carried out because of peptic ulcers is very low. On the other hand the number of urgent operations required by the complications--mainly by bleeding--of the ulcers is still rather high. AIM: Summarizing the possible treatments of these patients. METHODS AND DISCUSSION: The therapeutic endoscopy accompanied by the medical treatment has the primary role in the treatment of bleeding peptic ulcers. If this treatment fails, the surgical treatment is mandatory. Also surgical treatment is necessary in case of massive rebleeding. These kinds of operations are accompanied by a high morbidity and mortality, so it is understandable that the surgeons are trying to avoid them. After spontaneous or successful endoscopic cessation of bleeding in some cases rebleeding can be expected. The probability of rebleeding can be predicted with the consideration of predictive factors. With the help of such prognosis we can decide the indication of "early elective operation". In the surgical treatment we can expect better results with more "aggressive" type of operations (resections). The delay of an indicated operation, or repeatedly carried out endoscopic and medical treatment can reduce the survival chance of the bleeding patients. CONCLUSIONS: The treatment of upper gastrointestinal bleeding is a multidisciplinary task, which needs the correct cooperation of gastroenterologists, intensive therapists and surgeons.

Hospitalization for peptic ulcer bleeding: evaluation of a risk scoring system in clinical practice.

BACKGROUND: Upper gastrointestinal tract haemorrhage is a common cause of hospitalization: resource utilization in management of peptic ulcer bleeding varies considerably with no apparent effect on patient outcome. Several risk score systems based on endoscopic and clinical data have been proposed and validated in order to aid patient management. AIM: To assess clinical reliability of a scoring system and to define guidelines to improve efficiency of patient management without reducing efficacy METHODS: We considered all patients admitted to our unit for bleeding peptic ulcer over a one-year period. Every patient had an early endoscopy (within 12 hours) and therapy according to the appearance of the ulcer defined by Forrest classification. All subjects were classified into low-, intermediate- and high-risk patients on basis of clinical and endoscopic features according to "Cedar Sinai Medical Center predictive index" which was applied retrospectively in first six months then perspectively for the last period using the results obtained from first semester. For each risk group, we compared Length of Hospital Stay number of blood units used in transfusion, rebleeding rate, need for surgery as well as mortality in the two periods, using Student t test. We correlated Length of Hospital Stay and every score parameter by applying analysis of variance to results over the one-year period. RESULTS: Study population consists of 91 patients. Recurrent bleeding was observed in only three entering the high-risk group, only one of whom needed surgery Overall mortality was 9.8% (9 patients, only one for rebleeding). Variance analysis showed that the only parameter of the "Cedar Sinai Medical Center predictive index" which correlated with Length of Hospital Stay was comorbidity (p < or =0.05). Comparing the two periods, a close application of the score in the last six months allowed Length of Hospital Stay to be reduced in low-risk patients (t test with p=0.004) resulting in early discharge of 33% of cases without affecting patient outcome. CONCLUSIONS: This study confirms the reliability of the "Cedar Sinai Medical Center predictive index" in clinical practice improving the strategy of applying economic resources. Longer Length of Hospital Stay of intermediate- and high-risk groups is influenced more by comorbidities than by endoscopic findings. Early discharge was possible in one third of low risk patients. An accurate evaluation clinical para meters on admission together with early endoscopy may achieve the goal of reducing costs with a correct patient management.

[The surgical procedure in hemorrhages from acute ulcers and erosions in the gastroduodenal area]. / Khirurgicheskaia taktika pri krovotecheniiakh iz ostrykh iazv i érozii gastroduodenal'noi zony.

The experience obtained in treatment of 132 patients with acute ulcers and erosions of gastroduodenal area (GDA) is presented. Based on the analysis of surgical treatment in 38 patients with acute ulcers of the stomach and the duodenum, complicated with bleeding, the necessary scope of diagnostic modalities was suggested and substantiated, the terms and extent of surgical intervention were determined. The criteria for determining surgical policy in bleeding from acute ulcers and erosions of GDA were based on the intensity of blood loss and the endoscopical characteristics of acute ulcers (size, multiplicity, location). The necessity of computer pH-metry for optimization of the extent and method of surgical option in patients with acute ulcers and erosions of GDA is emphasized. In early postoperative period the authors have applied the method of external decompression of the stomach and the duodenum with enteral probe feeding. The results of treatment in all operated patients have been studied. Postoperative lethality made up 5.3%.

Non-variceal upper gastrointestinal bleeding and Forrest's classification: diagnostic agreement between endoscopists from the same area.

BACKGROUND AND STUDY AIMS: The lack of uniformity in defining the stigmata of hemorrhage in patients with bleeding ulcers is suggested by the wide range among published studies in prevalence and rebleeding rates for the same stigmata. Moreover there is, in published trials of endoscopic hemostasis, little standardization of definitions of stigmata of hemorrhage. The aim of this study was to assess the interobserver agreement among endoscopists from the same area (Piedmont and Valley of Aosta). PATIENTS AND METHODS: A workshop for 47 expert endoscopists was organized in order to evaluate their agreement in the diagnosis of stigmata of recent hemorrhage, according to Forrest's classification. During the meeting 25 videotapes from endoscopic examinations of patients with recent non-variceal bleeding were shown to the 47 endoscopists, who were asked to classify every lesion. RESULTS: The overall and beyond chance interobserver agreement was calculated by means of the kappa statistic. The overall agreement among endoscopists was highly significant (p < 0.001, kappa=0.60), while the beyond chance agreement varied from excellent to good for lesions with active bleeding (kappa=0.76 and kappa=0.61 for FIA and FIB lesions respectively), whereas for lesions with stigmata of recent hemorrhage kappa varied from 0.44 to 0.49. CONCLUSIONS: These data suggest the need for better knowledge of endoscopic criteria, in order to evaluate the results of endoscopic therapy and to assess new treatments.

[Comparative evaluation of endoscopic hemostatic methods in gastroduodenal ulcer hemorrhage]. / Sravnitel'naia otsenka sposobov éndoskopicheskogo gemostaza pri gastroduodenal'nykh iazvennykh krovotecheniiakh.

An experience with using various methods of endoscopic hemostasis for bleeding gastroduodenal ulcers is presented. The efficiency of different methods of arresting the bleedings depending on localization of the ulcer, place of the bleeding vessel and activity of the hemorrhage is discusses. The frequency of hemorrhages was shown to depend of the phase of the ulcer disease. A conclusion is made about high efficiency of using the combined endoscopic hemostasis as compared with using each of the methods of arresting the hemorrhage separately.

[The diagnosis of the severity of blood loss and general principles for the combined treatment of hemorrhage in relation to its severity]. / Diagnostike tiazhesti krovopoteria i obshchie printsipy kompleksnogo lecheniia krovotecheniia v zavisimosti ot ego tiazhesti.

The method of the blood loss severity estimation, basing on electrochemically registered globular volume deficiency (ECRGVD) was proposed. Classification of an acute blood loss severity accordingly to ECRGVD value was elaborated, general principles of complex treatment of hemorrhage were proposed.

Interobserver agreement on endoscopic diagnosis of bleeding peptic ulcers.

BACKGROUND: The aim of this prospective study was to evaluate the interobserver agreement of stigmata of recent hemorrhage of bleeding peptic ulcers. METHODS: Sixty-one consecutive adult patients were enrolled in the study and nine (three junior and six senior) endoscopists reviewed standardized video recordings of endoscopic examinations. Interobserver agreement was evaluated using the kappa (kappa) index, intraclass correlation coefficient, and proportion of agreement. Observer bias and poorly trained observers were investigated. RESULTS: Interobserver agreement was very good for oozing (kappa = 0.68), good for clot (kappa = 0.51), poor for spurting (kappa = 0.29) and visible vessels (kappa = 0.33), and excellent for the absence of stigmata (kappa = 0.82). Observer bias sometimes occurred and the number of poorly trained observers was low. The kappa indexes were significantly better in senior than in junior investigators: 0.48 +/- 0.16 versus 0.37 +/- 0.26, respectively, p < 0.05. The agreement between the in vivo evaluation and video tape recordings (intraobserver agreement) was good (kappa = 0.60 +/- 0.19). There was no training phenomenon between the first and the second half of the patient group. CONCLUSIONS: The endoscopic classification of bleeding ulcers might be simplified by limiting grading to a few classes. Special attention should be paid to the training of endoscopists.