This study investigates whether aspirin injury to the human gastric mucosa can be prevented by profound acid suppression with omeprazole, in a randomised, double blind, crossover design according to latin square. It was concluded that profound acid suppression can prevent aspirin induced gastric mucosal injury in normal subjects. This approach may prevent the development of peptic ulcers and their complications in patients taking aspirin and other non-steroidal anti-inflammatory drugs. Aspirin and non-steroidal anti-inflammatory drugs are strongly associated with peptic ulcer complications in the elderly in Britain. Assuming that the association may be causative, a number of approaches have tried to reduce the extent of damage caused by these agents.6 Although old people appear to be at greatest risk, it is difficult to evaluate possible therapeutic manoeuvres in this population. We have therefore investigated healthy adult volunteers whose acute responses appear to reflect those of older patients.7 Our previous studies show that acid inhibition by ranitidine8 and famotidine9 results in a reduction in gastric mucosal damage, as quantified by the rate of gastric mucosal bleeding and endoscopic appearance, but aspirin induced damage was not totally abolished. In the present study we have examined the hypothesis that gastric acid is necessary for the occurrence of aspirin induced gastric damage. To achieve virtual gastric anacidity we used omeprazole, an irreversible inhibitor of the proton pump in gastric parietal cells.'0 Methods bd for the last 48 h; (c) omeprazole 20 mg each morning for seven days plus aspirin 900 mg bd for the last 48 h; and (d) omeprazole 40 mg bd for seven days plus aspirin 900 mg bd for the last 48 h. The last doses of omeprazole and aspirin (or the corresponding placebo) were taken at 07 00 h and 07 30 h that is, 120 and 90 minutes, respectively, before measurement of gastric blood loss. Each of the above regimens was separated by a seven day washout period. The order of treatments was randomised according to a latin square design and the study conducted in double blind manner.

OBJECTIVE:To investigate the possible therapeutic role of omeprazole, a powerful proton pump inhibitor, in unselected patients presenting with upper gastrointestinal bleeding.DESIGN:Double blind placebo controlled parallel group study. Active treatment was omeprazole 80 mg intravenously immediately, then three doses of 40 mg intravenously at eight hourly intervals, then 40 mg orally at 12 hourly intervals. Treatment was started within 12 hours of admission and given for four days or until surgery, discharge, or death.SETTING:The medical wards of University and City Hospitals, Nottingham.SUBJECTS:1147 consecutive patients aged 18 years or more admitted over 40 months with acute upper gastrointestinal bleeding.MAIN OUTCOME MEASURES:Mortality from all causes; rate of rebleeding, transfusion requirements, and operation rate; effect of treatment on endoscopic appearances at initial endoscopy.RESULTS:Of 1147 patients included in the intention to treat analysis, 569 received placebo and 578 omeprazole. No significant differences were found between the placebo and omeprazole groups for rates of transfusion (302 (53%) placebo v 298 (52%) omeprazole), rebleeding (100 (18%) v 85 (15%)), operation (63 (11%) v 62 (11%)), and death (30 (5.3%) v 40 (6.9%)). However, there was an unexpected but significant reduction in endoscopic signs of upper gastrointestinal bleeding in patients treated with omeprazole compared with those treated with placebo (236 (45%) placebo v 176 (33%) omeprazole; p less than 0.0001).CONCLUSIONS:Omeprazole failed to reduce mortality, rebleeding, or transfusion requirements, although the reduction in endoscopic signs of bleeding suggests that inhibition of acid may be capable of influencing intragastric bleeding. Our data do not justify the routine use of acid inhibiting drugs in the management of haematemesis and melaena.

The effect of fish oil on the course of ulcerative colitis was investigated in a randomised blinded controlled study. Eighty seven patients received supplements of 20 ml HiEPA fish oil as triglyceride (4.5 g of eicosapentaenoic acid) or olive oil placebo daily for one year. The oils were given in addition to standard drug therapy and trial entry was stratified for disease activity. Fish oil significantly increased the eicosapentaenoic acid content of rectal mucosa to 3.2% of total fatty acids at six months, compared with 0.63% for patients on olive oil. This was associated with increased synthesis of leukotriene B5, and 53% suppression of leukotriene B4 synthesis by ionophore--stimulated neutrophils. Leukotriene B4 suppression persisted for at least two months after treatment was stopped. Treatment with fish oil resulted in measurable, but only limited clinical benefit. For patients entering the trial in relapse (n = 53), there was a significant reduction in corticosteroid requirement after one and two months treatment. There was a trend towards achieving remission (off corticosteroids) faster in the patients on fish oil, although differences were not significant. For patients in remission at trial entry or during the trial (n = 69), there was no significant difference in the rate of relapse by log rank analysis. We conclude that fish oil supplementation produces a modest corticosteroid sparing effect in active disease, but there is no benefit in maintenance therapy.

Roxatidine acetate is a new histamine H-2-antagonist of about twice the potency of ranitidine on a weight-for-weight basis. Two hundred and thirty-two patients participated in a double-blind randomized trial of duodenal ulcer healing comparing 300 mg ranitidine nocte with 150 mg roxatidine nocte. Endoscopy was repeated fortnightly to 4 weeks in each of four participating centres. Usual exclusion criteria applied but NSAID users were allowed. There were no important demographic differences between treatment recipients. Three analyses were used: protocol (dropouts and violators not included), intention-to-treat I (dropouts considered failures), and intention-to-treat II (dropouts considered failures, but violators outcome included). Healing rates differed markedly (but not significantly) with each analysis. After 2 weeks of treatment ulcers had healed in 51% versus 45% using the intention to treat I analysis with roxatidine and ranitidine, respectively; by the protocol analysis the healing proportions were 60% and 55%. These differences between treatments were not significant. After 4 weeks of treatment healing rates ranged from 71% to 83% on roxatidine and between 69% and 84% on ranitidine depending on the analysis. Differential healing proportions of smokers and non-smokers were non-significant (83% vs. 79%). Both drugs were well tolerated and adverse events were similar with each agent. Roxatidine should prove as effective as ranitidine for acute duodenal ulcer treatment.

1. An increasing body of data suggests that the antihaemostatic as well as the ulcerogenic actions of aspirin and other non-steroidal anti-inflammatory drugs may be operative when patients present with haematemesis and melaena. 2. We therefore developed methods to allow separate evaluation of the erosive and anti-haemostatic actions of aspirin in the human gastric mucosa. Volunteer subjects took 300 mg of aspirin daily in the morning or 600 mg of aspirin four times a day for 5 days under blinded randomized conditions. Changes in spontaneous gastric microbleeding, endoscopic signs of injury, spontaneous bleeding per gastric erosion, biopsy-induced bleeding and eicosanoids were studied. 3. Both doses of aspirin significantly inhibited gastric mucosal synthesis of prostaglandin E2 and reduced the serum thromboxane concentration. Erosions developed and regressed rapidly; compared with baseline 300 mg of aspirin daily in the morning caused substantial numbers of gastric erosions to develop (mean 5.3, 95% confidence limits 2.7-10.2) but this was significantly less than that caused by 600 mg of aspirin four times a day (10.9, 7.2-16.5, P less than 0.05). The presence of erosions was associated with enhanced spontaneous bleeding, but only during aspirin administration. 4. Aspirin significantly increased bleeding induced by mucosal biopsy and was associated with significant enhancements in the rate of bleeding per gastric erosion. Bleeding rate per erosion but not biopsy-induced bleeding showed a significant dose-related increase with 600 mg of aspirin four times a day. Enteric coating reduced endoscopic signs of injury, but did not affect the impaired haemostasis caused by aspirin.(ABSTRACT TRUNCATED AT 250 WORDS)

They question whether we were able to empty the balloon between inflations. Because we measured the volume inserted into the balloon on each occasion and also the volume recovered we were able to ensure that accumulation of fluid in the balloon did not occur. They criticise the use of a fluid filled catheter to detect intraballoon pressure (standard practice by most other workers). We agree that different absolute values could be obtained using different catheters but find it difficult to see how the varying inter and intraindividual results we obtained could be the result of using different catheters. They also question whether our normal subject group, all ofwhom were symptom free and without past or current history of bowel symptoms, could have inadvertently included subjects with irritable bowel syndrome. If it is possible for an adult who feels healthy and who is not aware of any symptoms referable to the gastrointestinal tract to be suffering from irritable bowel syndrome then we have to admit guilt, but then anybody studying normal human physiology of the gut will commit the same mistake. Varma and Smith also make further points concerning irritable bowel syndrome, suggesting that there may be subgroups who could perform differently. We would not disagree with this possibility and indeed two of our patients with predominant diarrhoea (cases 4 and 24) had maximal tolerable volumes to distension which were at the bottom of the patient range. The point that we made in our paper, however, is that neither of these two patients fell outside the 95% confidence limits for our normal range and while there may indeed be a difference between irritable bowel syndrome patients who are at the extremes of symptoms, neither of these two extremes can be regarded as being 'abnormal.' They question the past history of the patients and their age: we also considered this and mentioned it in our discussion. It seems likely that failure to control for differences between age and sex of the two groups would have increased the differences between the groups rather than reduced them. They comment that rectal volumes tolerated by subjects would vary with the confidence and experience of the subjects. We wholeheartedly agree with this statement and indeed this was one of the aims of our study. Our results showed that variability progressively reduced in an individual as the study continued. This is discussed in detail in our paper. They go on to mention that individual variability would not be so great if other sensory or motor indices of rectal distension were used. The major difficulty identified and discussed in our paper is that the proctometrogram as traditionally employed relies on sensory end points, which are subject to individual interpretation rather than independent measurement. They comment that the tracings which we published of the pressure-volume relations seem to be consistent in each individual and indeed we would agree with this. Our point again, however, is that it is the sensory end points that are variable, even though the pressure-volume curve is consistent. The major problem in analysis of pressure-volume curves, however, is to adequately define in mathematical terms the pressure-volume curve itself. It is of doubtful use to talk about compliance (as do many workers in the field), since as is evident from all published data, the slope of the pressure-volume curve varies during inflation so that there are a number of compliancies rather than a single value. Sun et al also criticise our selection of patients and suggest that our sample was biased towards constipation. Thirteen of 26 patients were constipated, two were predominately diarrhoea sufferers, five had both diarrhoea and constipation, the rest were unaffected by major alterations in stool consistency. These patients therefore do not seem too different from those reported in other recent studies of irritable bowel syndrome such as that of Prior et al (Sun et al, reference 1) in which 27 of 55 irritable bowel patients were constipated. We would not argue with their suggestion that patients with diarrhoea predominant symptoms may differ from those with constipation predominant symptoms and also are aware that those individuals with diarrhoea in our study had the lowest tolerated volumes of the groups. We point out again, however, that none of these individuals fell outside the range of normality as defined by our volunteer data, so that while proctometrography may differentiate between irritable bowel syndrome patients with diarrhoea and constipation, it is difficult to see how the technique could be used to distinguish the irritable bowel syndrome from normal. From the interest which our article has stimulated it is evident that more data are required to understand the basic factors influencing pressure-volume relations in the rectum. We look forward to seeing more data from both correspondents to help clarify these difficulties. D G THOMPSON Department ofMedicine, Section ofGastroenterology, Hope Hospital, Eccles Old Road, ManchesterM6 8HD

A postal questionnaire inquiring about routine sedation and premedication practice for upper gastrointestinal endoscopy was sent to 1048 doctors. Of 665 appropriate returns, 81% were from consultant physicians and surgeons. Most endoscopists (90%) reported using an intravenous benzodiazepine for at least three quarters of endoscopies and 54% of physicians and 69% of surgeons always did so. Midazolam was the intravenous sedative used by a third of all respondents and 13% also used an additional intravenous agent, usually pethidine. Over the previous two years a total of 119 respiratory arrests, 37 cardiac arrests, and 52 deaths were identified. Adverse outcomes were reported more frequently by consultant physicians, by those who 'titrated' the intravenous sedative, and by those who used an additional intravenous agent, but were reported equally frequently by endoscopists using midazolam and endoscopists using diazepam. There is an urgent need for a prospective study to identify the circumstances and risk factors associated with adverse outcomes related to endoscopy.

SUMMARYTo assess the effect of indomethacin on gastric acidity and to identify a potential pharmacodynamic interaction between indomethacin and ranitidine, we measured nocturnal acidity on half‐hourly aliquots of gastric contents from 10 volunteers on the seveneth day of four dosing regimens given in a randomized double‐blind manner. These were indomethacin (50 mg t.d.s.) and ranitidine (300 mg in the evening) together or alone with matching placebos. Median nocturnal acidity on placebo was 41.7 mmol/L (range 67.6–25.1 mmol/L) and was 39.8 mmol/L (63.1–24.0 mmol/L) on indomethacin (N.S.). During ranitidine dosing it was 0.4 mmol/L (21.3–0.0 mmol/L) without and 0.8 mmol/L (43.7–0.0 mmol/L) with concurrent indomethacin, representing 99 and 98% decreases in gastric acidity (P< 0.01) compared with placebo. Indomethacin did not increase overnight gastric acidity and did not influence the suppression of acidity produced by ranitidine. It is unlikely that the ulcerogenic potential of indomethacin is explicable by an effect on gastric acidity.

OBJECTIVE:To evaluate the medical impact of reactive pharmacy intervention.DESIGN:Analysis of all interventions during 28 days by all 35 pharmacists in hospitals in Nottingham.SETTING:All (six) hospitals in the Nottingham health authority (a teaching district), representing 2530 mainly acute beds, 781 mental illness beds, and 633 mainly health care of the elderly beds.PATIENTS:Hospital inpatients and outpatients.INTERVENTIONS:Recording of every important intervention made by pharmacists to prescriptions for both inpatients and outpatients when they perceived inadequacies of drug prescription or administration, including characterisation of the problem, coding of outcome, recording of time taken to initiate and resolve intervention, and grade of prescribing doctor. The problems were independently assessed for their potential to cause medical harm.RESULTS:769 Interventions (about 2.9% of prescriptions) were made, of which 60 concerned prescriptions rated as having a major potential for medical harm. The commonest problems concerned dosage, which was wrong in 280 prescriptions (102 for antibiotics) and not stated in 50 (one for antibiotics), especially those associated with a major potential for medical harm (32 prescriptions). These concerned sedatives; analgesics; cardiovascular drugs or diuretics; and iron, vitamin, or mineral preparations. Also common were overprolonged prescription of antibiotics (48 prescriptions), confusion of drug names (nine), and inadvertent coprescription of excessive quantities of aspirin or paracetamol in plain and compound preparations (seven). The pharmacist's recommendation was accepted in 639 instances (86%), and the prescription was altered in 575, leading to an appreciable (246 cases) or minor (231 cases) improvement. Interventions had little effect on costs; 427/646 had no effect and 130 produced savings less than 50p. Pharmacy intervention (730/769 interventions) occupied on average 41 minutes per pharmacist per week.CONCLUSIONS:Most reactive pharmacy interventions concerned prescribing errors with a limited potential for medical harm, but a small number of detected errors with a major potential for medical harm; cost savings were not appreciable.

1. We investigated the effect of the haemostatic agent ethamsylate on aspirin-induced gastric mucosal bleeding. 2. Eighteen healthy subjects were studied three times: at the end of 48 h periods of treatment with (a) placebo, (b) aspirin 600 mg four times daily, (9 doses) and (c) aspirin 600 mg four times daily with each dose preceded by ethamsylate 500 mg. 3. At the end of each treatment period gastric mucosal bleeding into timed gastric washings was quantified using the orthotolidine reaction. 4. Aspirin increased bleeding from a rate on placebo of 1.2 microliters 10 min-1 geometric mean (95% confidence limits) (0.7-1.8) microliters 10 min-1 to 20.0 (11.6-34.2) microliters 10 min-1, (P less than 0.01). The rate of bleeding after aspirin preceded by ethamsylate [14.1 (8.5-23.4) microliters 10 min-1] was not significantly different from that after aspirin alone. 5. We conclude that ethamsylate does not reduce acute aspirin-induced gastric mucosal bleeding in healthy humans.