A30 AGA ABSTRACTS
G0121 APPLICATION OF SPEECH TECHNOLOGY IN THE ENDOSCOPIC LABORATORY FOR REMOTE MACHINE CONTROL AND REPORTING. B. Moln~r *J. Gergely, K. Papik, *G. T6th, L. Pr6nai, Z. Tulassay, Clin. Gastroent. Unit, 2 no. Dept. of Med., Semmelweis Univ., *Gabor D. Tech. High School Budapest, Hungary Aims: The application of speech technology in the medical daily routine would allow the increase of efficiancy and decrease of personal costs. A simulation program was developed for reporting gastroscopy investigation result and control of laboratory equipments using speech as input and output interface. Methods: Functions of the endoscopy laboratory equipments were described as multivariate parameter system. The core unit is a central speech computer, which evaluates the doctor's command, compares it to the status information obtained from the specific worktool and performs a spoken answer. We developed special programms for the grammatical analysis of command sentences and definition of machine answers. The speech recognition functions were obtained from the Dragon VoiceTools speech application development library (Dragon Co., Ma. USA), running on a the Soundblaster 16 ASP (Creative Labs, Singapore). The speech synthesis was performed with the PC-Robot programm (Nikol. Co. Hungary). We used a wirelesss microphone for speech control. Resu!ts: In our database, we had sentences for commanding the endoscopy equipments and altogether 100 unidentified patient records. The patient report's sentences were grouped according to keywords which describe the major part of a gastrosendoscopy report. They were the: endoscop. eosephagus, cardia, fundus, corpus, antrum, pylorus, bulbus, postbulbar section and the pathological major findings: erosion, ulceration, malignancy furthermore biopsy and diagnosis. Speech samples were collected from 3 male low-tone investigators. We had alltogether 3546 words in our vocabulary. The training of the entire vocabulary took 4-6 hours for one person. The personal adaptation of the vocabulary took less than 10 minutes training. The recognition rate was 95%. Using a 200 Mhz Pentium computer, the speech-to-text conversation for one word with 12 alphabets was under 1 second. The users of the system found the speech back up relatively uncomfortable and slow, however, they were free from looking on a monitor. The dialogue with a black box system via wireless communication, activates human interest for finding the technological edge of the system, which decreases after some experimental sessions with the dialogue box. Conclusions; Speech recognition is fast and safe enough for routine applications. The development of a speech based dialogue needs detailed analysis and description of man-machine interaction to loose the redundant informations. • G0122 PROPHYLACTIC ADMINISTRATION OF ANTIBIOTICS BEFORE INTERVENTIONAL ERCP IN PATIENTS WITHOUT PREEXISTING INFECTION. M. Mrol31, S. Kiipferling l, M. Ortner l, H. Briedigkeit2, U. Goebel 2, H. Lochs I 1IV. Medical Department Charitt, Humboldt-Universitiy Berlin, 2Institute of Microbiology Charitt, Humboldt-University Berlin The aim of this study was to determine the effect of prophylactic administration of antibiotics before ERCP. Furthermore, we compared usual indicators for a bacterial infection with serum levels of procalcitonin and IL-6 which have been shown to increase during an infection. Methods: Between August 1996 and August 1997, 89 patients (45 men and 44 women) treated at our hospital were prospectively randomized in 2 study groups. Patients with signs of infection before ERCP were excluded. 42 patients received intravenous ceftriaxon (Rocephin®, 2 g) 2 hours before ERCP (group A). 47 patients did not receive antibiotics (group C). Infection for defined as either positive blood cultures and/or at least 2 of the following parameters: CrP, WBC, procalcitonin, IL-6 which were determined before, 4 and 24 hours after intervention. Results: In the 89 patients 130 interventions (sphincterotomy, endoprothesis, stone extraction, cholangioscopy, brush cytology, biopsy, photodynamie therapy, dilatation) were performed during ERCP, without significant differences between group A and C. In group C 7/47 patients (14.8%) developed elevated blood markers and clinical signs of bacterial infection after the intervention. Of the patients in group A who received antibiotic prophylaxis only 2 patients (4.8%) exhibited signs of a systemic bacterial infection (p > 0.05). Conclusions: Prophylactic administration of 2 g of ceftriaxon 2 hours before interventional ERCP reduced the absolute number of systemic bacterial infections but failed to reach statistical significance. A prospective, randomized trial involving a larger cohort of patients will be necessary to confirm our results.
GASTROENTEROLOGY Vol. 114, No. 4 G0123
SAFETY AND EFFICACY OF LANSOPRAZOLE: RESULTS OF A PROSPECTIVE FOLLOW-UP STUDY IN DAILY PRACTICE. P. Miillerl, B. Simon2, Krankenhaus Salem, Heidelberg l, FRG, Kreiskrankenhaus Schwetzingen, Schwetzingen2, FRG Aim: As a part of the postmarketing surveilliance program, a large prospective follow-up study was designed to determine the safety and efficacy of this second generation proton pump inhibitor. Method: Patients presenting with gastric (GU) or duodenal (DU) ulcer disease or reflux-disease (GERD) have been enrolled between January 1996 and April 1997 into this study. The usual dosage of lansoprazole was 30 mg or 60 mg once daily. Results:
male (%) female (%) age (mean ± SD) smokers (%) recurrent disease (%) duration of treatment (days) H.p.diagnosis done H.p.-positive (e/o) eradication rates in H.p.-positive patients after eradication therapy (%) relief of pain at the end of therapy (%) healing rates (%)
GU
DU
GERD
14056 57,6 40,8 51 ± 15 47,1 39,1 21+_.11 9826 58,7 92,1
23455 60,5 37,6 48 ± 14 52,1 38,5 19_+9 16186 64 93,0
25179 52,3 45,6 52 -" 15 41,2 29,7 23-.13 13924 28,3 89,5
96,3
97,6
95,5
89,9
93
79,1
Only 311 patients (0,46%) reported side effects; main symptoms were nausea (0,05%), diarrhoea (0,07%), vertigo (0,02%) and headache (0,03%). Conclusion: Effectiveness appeared to be comparable to results found in clinical trials for lansoprazole. Only in 0,46% o f > 60.000 patients side effects were reported. Lansoprazole was found to be safe in this naturally occuring group of patients. This work was funded by Takeda Pharma, Germany • G0124 HOSPITAL USE OF PROTON PUMP INHIBITORS: A COMPLETED AUDIT OF THE INTRODUCTION OF A PRESCRIBING PROTOCOL ON IN-PATIENT PRESCRIPTIONS. M Newton, G Smith, WR Burnham, C Nicholls, B So. Havering Hospitals' Trust, Romford, Essex, UK. Background: Proton pump inhibitors (PPIs) are highly effective for appropriate upper gastrointestinal disorders, but also very costly. The aim of this audit was to review the effect of guidelines for PPI use introduced for hospital prescribing. Methods: During a one month period all prescriptions for PPIs in hospital inpatients (except obstetrics and paediatrics) were identified and the case notes reviewed by a consultant gastroenterologist. Guidelines for the prescription of PPIs based upon drug data sheets were approved by the hospital's Drugs and Therapeutics' Committee. Pharmacists were asked to query any prescription which fell outside the recommended guidelines. Subsequently a second 30 day audit was carried as before. Prescription of a PPI was regarded as definitely indicated if within the guidelines. Those outside the guidelines were judged on clinical grounds to be a reasonable use of PPIs or to be definitely not indicated. Results: During the first 30 day audit period 105 patients (7% of total inpatients in this period) were prescribed a PPI, of whom notes were retrieved in 102. During the second audit, 99 patients (6% of in-patients) were prescribed PPIs and all notes retrieved. The proportion of patients who started a new prescription during the current admission (32/102 v 16/99) and on long-term therapy ( > 6 months) (30/102 v 36/99) did not change. 77% of PPI prescriptions had originated in hospital. A significant reduction in the use of higher dose PPI was seen (28/102 v 6/99, p < 0.004). During the first audit period 94 patients were prescribed omeprazole but 52 had lansoprazole during the second audit. At least 25% of patients had had no prior investigation of the upper GI tract (73 v 49; p=0.06). There was no indication for the use of PPIs in one quarter of prescriptions and this appeared to increase following guideline introduction (27 v 38; p=0.07). Of the remainder less than half (51/102 v 29/99; p=0.003) were within the guidelines. Condusion: One in 17 hospital in-patients were prescribed a proton pump inhibitor. The majority of prescriptions had been hospital initiated. Despite the fact that one quarter of these prescriptions were not within guidelines, their introduction did not affect the number of PPI prescriptions but did reduce the dose of drug prescribed. Use of guidelines may affect drug expenditure by encouraging more appropriate dosing but they had no influence on the reasons for PPI use.