course for administration is not stated. During the procedure the patient suffered a cardiac arrest. The prompt administration of the specific benzodiazepine antagonist flumazenil resulted in rapid recovery. Sixteen hours later the patient died after a gradual deterioration. Further measures taken to stop this decline are not detailed. We are then informed that the cause of death was cerebral anoxia due to incomplete reversal of midazolam induced respiratory depression. Firm evidence to support this conclusion seems to be lacking. Indeed, if flumazenil had not been available there is a good chance that the patient would have died sooner. Elderly patients show increased sensitivity to the effects of benzodiazepines.2 In such patients 4 mg of midazolam together with the passage of an endoscope can cause important respiratory obstruction.' The resultant hypoxia can be sufficient to cause collapse and arrhythmias4 and increase the risk of aspiration.5 In many instances these problems can be avoided by the simple expedient of giving all elderly and unfit patients supplemental oxygen.6 Dr Lim states in support of his claim a paper by Mora et al.7 This study comprised just 10 patients who received large doses of diazepam (up to 210 mg) over periods as long as three and a half hours. Only five of the patients had important respiratory depression, and, of these, four showed improvement after receiving flumazenil. Pharmacokinetic data28 lend no support to Dr Lim's hypothesis. The duration of drug action after administration of a single dose depends on drug distribution rather than clearance or elimination.8 It is highly unlikely that appreciable concentrations of midazolam were present at the time of death unless the patient had pronounced hepatic impairment. It is more likely that a combination of
factors-that is, pre-existing disease (short and long term) coupled with the physiological insult of a cardiac arrest -were responsible for the patient's death. It is entirely inadequate to claim on the basis of this one case that flumazenil does not reverse midazolam induced respiratory depression. Furthermore, there is no published evidence that
advocates resedating patients after they have received flumazenil if single low doses of short acting benzodiazepines have been given. BRIAN R P BIRCH RONALD A MILLER
Departments of Minimally Invasive Surgery, Institute of Urology and Royal Northern Hospital, London I Lim AG. Death after flumazenil. Br Med J 1989;299:858-9.
(30 September.) 2 Harper KW, Collier PS, Dundee JW, Elliott P, Halliday NJ, Lowry KG. Age and nature of operation influence the pharmacokinetics of midazolam. BrJ7 Anaesth 1985;57:866-7 1. 3 Bell GD, Reeve PA, Moshiri M, et al. Intravenous midazolam: a study of the degree of oxygen desaturation occurring during upper gastrointestinal endoscopy. BrJI Clin Pharmacol 1987; 23:703-8. 4 Rostykus PS, McDonald GB, Albert RK. Upper intestinal endoscopy induces hypoxaemia in patients with obstructive pulmonary disease. Gastroenterology 1980;78:488-91.
5 Prout BJ, Metreweli C. Pulmonary aspiration after fibre-endoscopy of the upper gastrointestinal tract. Br Med J 1972;iv: 269-71. 6 Bell GD, Morden A, Carter A, Coady T. Midazolam antagonism. Lancet 1988;ii:388. 7 Mora CT, Toriman M, White PF. Effects of diazepam and flumazenil on sedation and hypoxic ventilatory response. Anesth Analg 1989;68:473-8. 8 Greenblatt DJ, Abernethy DR, Locniskar A, Harmatz JS, Limiuco RA, Shader R. Effect of age, gender and obesity on midazolam kinetics. Anesthesiology 1984;61:27-35.
Hearing problems of elderly people SIR,-With regard to Dr Gordon Hickish's leading article' and the reply from Dr John R Hughes,2 I would like to clarify the situation. As the population ages and is exposed to more industrial
and leisure noise the requirements for hearing aids will increase. If an audit shows that there are large numbers of patients who may benefit from a hearing aid and that the current resources in ear, nose, and throat departments are inadequate to cope (as suggested by Dr R W R Farrell and colleagues3) the situation needs to be radically rethought. Elderly people who require spectacles are able to obtain these without waiting for months for an outpatient appointment and without having to attend at the convenience of the health service at district hospitals. It must therefore be possible to design a method whereby hearing aids could be prescribed by general practitioners and fitted for the vast majority of the population without the need for the consultant service. Consultants should then be seen by those people who are appropriately referred after having been assessed either in the practice or at a hearing aid centre, which would enable specialists to use their high degree of training for those patients in whom it is
appropriate. With regard to Dr Hughes's letter, perhaps the reason that some patients do not use their hearing aids appropriately is that they have never been adequately instructed in their use or that they find that the hearing aids provided are not acceptable. If a service was introduced in which the users could have some direct contact with the suppliers the development of cheaper and more effective hearing aids might be possible. P D SPRACKLING
Derby Road Health Centre, Nottingham NG7 IQG I Hickish G. Hearing problems of elderly people. Br Med J
1989;299:1415-6. (9 December.) 2 Hughes J. Hearing problems of elderly people. Br Med J 1990;300:122. (13 January.)
3 Farrell RWR, Parker A, Buffin JT. Hearing problems of elderly people. BrMedJ 1990;300:122. (13 January.)
Selective decontamination of the digestive tract SIR,-We deduce from the correspondence'4 following Dr Sanderson's leading article that Dr Sanderson has altered the main reason for his sceptical subtitle. Originally he suggested that mortality had to be the ultimate criterion, but now he contends that further evidence is required for the need of the systemic component as well as for the roles of oropharyngeal and gastrointestinal decontamination and gastric acidity. The requirement for reduced mortality before a technique may be implemented in intensive care units is debatable. A consequence could be the abandonment of other treatments, such as stress ulcer prophylaxis, the use of Swan-Ganz catheters, total parenteral nutrition, and even the expensive conventional methods of preventing infection, none of which has been shown to reduce mortality significantly.6 Attempts have been made to evaluate the effect of selective decontamination on mortality in 12 studies, only four ofwhich were quoted.7 The fact that most studies showed little difference in mortality resulted from the fact that none had the proper design to establish a difference. A mortality study should be performed in a patient population in whom infection has been proved to contribute to death. In fact, selective decontamination has thrown new light on the interaction of acquired infections with outcome, suggesting that patients die with rather than of infections. A significant reduction in mortality is expected primarily in non-infected patients with curable disease.8 A significant reduction in the amount of systemic antibiotics used during selective decontamination has been shown,' and the method also reduces carriage of multiresistant bacteria.'0 The preven-
tion of severe infections has a major impact on costs of treatment and daily management of the critically ill. Although these issues need further clarification, each of them may justify implementation. The correspondence has not elucidated the need for a systemic antibiotic. This has two purposes: firstly, to eradicate potential bacterial pathogens, such as pneumococci, commonly present in the oropharynx and not affected by the non-absorbable agents; and, secondly, to prevent the colonisation of the lower airways by Gram negative bacteria." The choice of cefotaxime has been further supported in a large multicentre trial in which the percentage of Gram negative bacilli was consistently higher (>50%) than the percentage of Gram positive bacteria implicated in the initial pneumonia.12 Dr Sanderson stated that "more individual patients would receive antibiotics under this regimen than if infections were treated as they arise," but in our experience there are few critically ill patients who do not receive antibiotics in the intensive care unit. Most receive systemic antibiotics for treating existing and acquired infections (more than 70% of the patients were infected on admission in the Glasgow study'), but the superinfection rate is also high, and the emergence of resistance is an intractable problem. Dr Sanderson mentions the role of gastric acidity. Although low gastric acidity might prevent lower airway infections by controlling overgrowth of gastric micro-organisms,'2 many critically ill patients already have a gastric pH higher than 4. In this type of patient the physiological gastric barrier does not function properly. The reduction in the incidence of pneumonia achieved by sucralfate"3 may be explained by its antibacterial activity rather than by its preservation of gastric acidity. Furthermore, sucralfate does not influence oropharyngeal carriage, which is important in the pathogenesis of pneumonia. 14 Moreover, the administration of non-absorbable antimicrobial agents into only the stomach did not reduce the incidence of pneumonia, but the combination of oropharyngeal and gastrointestinal decontamination was better than sucralfate in preventing
H K F VAN SAENE E M LEONARD A PERCIVAL
Department of Medical Microbiology, University of Liverpool, Liverpool L69 3BX C P STOUTENBEEK Onze Lieve Vrouwe Gasthuis,
Amsterdam, The Netherlands
I Ramsay G. Selective decontamination of the digestive tract. BrMedJ 1990;300:190. (20 January.) 2 Coad N. Selective decontamination of the digestive tract. BrMedJ 1990;300:190. (20 January.) 3 Thomas DR, Cole DS. Selective decontamination of the digestive tract. Br MedJ7 1990;300:190. (20 January.) 4 Sanderson PJ. Selective decontamination of the digestive trpct. BrMedJ 1990;300:190. (20 January.) 5 Sanderson PJ. Selective decontamination of the digestive tract. BrMedJ 1989;299:1413-4. (9 December.) 6 Lacroix J, Infante-Rivard C, Jenicek M, Cauthier M. Prophylaxis of upper gastro-intestinal bleeding in intensive care units: a meta-analysis. Crit Care Med 1989;17:862-9. 7 van Saene HKF, Stoutenbeek CP, Stoller JK. Selective decontamination of the digestive tract (SDD) in ICU: current status and future prospects. Crit Care Med (in press). 8 Gross PA, van Antwerpen C. Nosocomial infections and hospital deaths. A case-control study. Am J Med 1983;75: 658-62. 9 Stoutenbeek CP, van Saene HKF, Miranda DR, Zandetra DF, Binnendijk B. The prevention of superinfection in multiple trauma patients. J Antimicrob Chemother 1984;14(suppl B): 203-11. 10 Brun-Buisson C, Legrand P, Rauss A, et al. Intestinal decontamination for control of nosocomial multiresistant Gram-negative bacilli: study of an outbreak in an intensive care unit. Ann InternMed 1989;11O:873-81. 11 Langer M,,Mosconi P, Cigada M, Mandelli M, and ICU group of infection control. Longterm respiratory support and risk of pneumonia in critically ill patients. Am Rev Respir Dis 1989;140:302-5. 12 Maki DG. Risk factors for nosocomial infection in intensive care. "Devices versus nature" and goals for the next decade. Arch Intern Med 1989;149:30-5.
BMJ VOLUME 300
17 FEBRUARY 1990
13 Driks MR, Craven DE, Celli DR, et al. Nosocomial pneumonia in intubated patients given sucralfate as compared with antacids or histamine type 2 blockers. N Engl J Med 1987;317: 1376-82. 14 Flaherty J, Kabins SA, Weinstein RA. New approaches to the prevention of infection in intensive care unit patients. In: van Saene HKF, Stoutenbeek CP, Lewin P, McA Ledingham I, eds: Infection control by selective decontamination. Berlin: Springer Verlag, 1989:184-8. 15 Stoutenbeek CP, van Saene HKF, Miranda DR, Zandstra DF, Langrehr D. The effect of oropharyngeal decontamination using topical non absorbable antibiotics on the incidence of nosocomial respiratory tract infections in multiple trauma patients._ Trauma 1987;27:357-64.
SIR,-The discussion on the effect of selective decontamination on mortality in intensive care units' cannot be resolved by an uncritical look at data from currently published studies. Selective decontamination is not a miraculous cure for the critically ill, merely an effective prevention regimen for use in units in which serious nosocomial infection is frequently seen. Research evidence suggests that reductions in mortality as a result of using gastrointestinal decontamination and systemic cefotaxime (SPEAR) are likely to be modest. Regression analysis of data from the Glasgow study2 suggested that admission factors as measured by Apache II3 and sepsis4 scores were of greater importance in deciding outcome than infection acquired in the intensive therapy unit. Only a very large study would show whether a small but important improvement in mortality can be achieved by using selective decontamination. For example, the second international study of infarct survival (ISIS-2)' needed 17 187 patients with myocardial infarction to show a 25% reduction in mortality from streptokinase infusion from 12% to 92%. The ability of a trial to decide between selective decontamination and conventional management in terms of affecting overall mortality relies on the number of deaths rather than the number of patients entered.6 Any reduction in mortality as a result of selective decontamination will depend on the incidence of infection acquired in the unit and subsequent mortality. In the Glasgow trial only 10 of 161 controls died in intensive therapy units after acquiring infection in the unit. This represents only a quarter of the 39 deaths in the group. If we consider a hypothetical study comparing selective decontamination with conventional treatment in which the incidence of acquired infection in the controls remains at 24% in those staying over 48 hours, and the subsequent mortality is 50%, as was seen in our trial, the study would need 3200 admissions to observe 100 deaths in a control group of 1600 patients after an episode of acquired infection. We would also expect 300 deaths from other causes, giving a total of 400 deaths and 1200 survivors in the control group and a mortality of 25%. If half of the deaths after acquired infection were prevented by selective decontamination (a highly optimistic outcome), resulting in a total of 350 deaths and 1250 survivors in the treatment group, overall mortality would fall by only 3% to 22%. This would only just achieve a significant result by X2 testing (p=005, 95% confidence interval 6 06 to 0 19). Smaller reductions in mortality would need an even larger study to achieve significance. To conduct such a study would require the participation of several units, uniform admission and management procedures, and some agreement on the difficult issue of diagnosis of infection, particularly pneumonia. Mortality, however, is not the only criterion that should be looked at. Miranda et al have shown considerable financial savings as a result of selective decontamination of the digestive tract.8 We have also seen savings in the use of disposables such as paper gowns, intravenous administration sets, and parenteral nutrition that outweigh the relatively small increase in antibiotic costs. If there continues to be no appreciable problem with antibiotic resistance the present policy of using selective
decontamination in all ventilated patients who stay more than 24-48 hours in the intensive therapy unit could be justified based on the economic benefits alone. J C McDONALD
Department of Anaesthesia, Western Infirmary, Glasgow GIl 6NT 1 Sanderson PJ. Selective decontamination of the digestive tract: value in intensive care units not proved. Br MedJ7 1989;299: 1413-4. (9 December.) 2 Ledingham IMA, Alcock SR, Eastaway AT, McDonald JC, Ramsay G, Mackay I. The use of a triple regimen of selective decontamination of the digestive tract, systemic cefotaxime and microbial surveillance for prevention of acquired infection in intensive care. Lancet 1988;i:785-90. 3 Knaus WA, Draper EA, Wagner DP, el al. Apache II: a severity of disease classification system. Crt Care Med 1985;13:818-29. 4 Elbute EA, Stoner HB. The grading of sepsis. Br J Surg 1983;70:29-3 1. 5 Second International Study of Infarct Survival Collaborative Group. Randomised trial of intravenous streptokinase, oral aspirin, both or neither among 17 187 cases of suspected acute myocardial infarction: ISIS-2. Lancet 1988;ii:349-59. 6 Peto R, Pike MC, Armitage P, et al. Design and analysis of randomised clinical trials requiring prolonged observation of each patient. 1. Introduction and design. BrJ Cancer 1975;34: 585-612. 7 Boag JW, Haybittle JL, Fowler JF, Emery EW. The number of patients required in a clinical trial. BrJ Radiol 1971;44:122-5. 8 Miranda DR, van Sane HKF, Stoutenbeek CP, Zanstra DF. Environment and costs in surgical intensive care units. The implication of selective decontamination of the digestive tract. Acta Anaesthesiol Belg 1983;34:223-31.
Information please SIR,-Dr D Clements' presumably missed the letter in which Dr A J Robertson and I commented on the advantages of patients having a unique identification number for all general practice and hospital purposes.2 Experience in Tayside has confirmed the potential of such a system for medical audit and resource management. For example, linkage of events such as admissions of patients to different hospitals with different diagnoses (even though the underlying disease process may be the same) is easy if the same patient number is used for case records in each hospital. I Information on survival can also be incorporated as the community health index is updated by mortality data from the registrar general. A further advantage is that confidentiality is enhanced as the basis of linkage is the index number, and no names need be sent to the computer department or need appear on subsequent printouts. The first step for those south of the Tweed is to abolish the unfortunate separation of district health authorities and family practitioner committees and adopt the Scottish system of having primary care divisions as integral parts of health boards. N R WAUGH Tavside Health Board, Dundee DDl 9NL I Clements D. Information please. BrMed
January.) 2 Waugh NR, Robertson AJ. Evaluation of a call programme for cervical cytology screening. Br Med J 1989;299:855.
(30 September.) 3 Waugh NR. Amputation in diabetic patients-a review of rates, relative risks and resource use. Community Med 1988;10: 279-88.
Fibromuscular dysplasia of renal arteries and acute loin pain SIR,-The article by Dr Simon J Stinchcombe and colleagues reminds us of the importance of diagnosing arterial occlusion as a rare cause of acute loin pain.' They have, however, overestimated its incidence by placing it at the top of
the list of causes of acute loin pain with nonfunctioning kidney. Ureteric obstruction-the commonest cause-is placed last. We also have misgivings about their proposed method of investigating acute loin pain. Surely a full clinical assessment, urine analysis, and microscopy and plain film radiography should be the initial gambit rather than intravenous urography. Only those patients without evidence of acute pyelonephritis or radio-opaque ureteric calculus require emergency intravenous urography. Of the remainder, those with pyuria, bacteriuria, and fever require ultrasonography to exclude pyonephrosis which would need drainage, and those with a radio-opaque calculus on plain film radiography should be investigated electively, which may include intravenous urography. The authors do not suggest further investigation of patients with a hydronephrotic, non-functioning kidney. An antegrade pyelogram would not only confirm the presence of obstruction but would also show the level and nature of the obstructing lesion. Finding a normal pelvicaliceal system on ultrasonography does not exclude an obstructive cause. The only further investigation in such patients suggested in the authors' flow chart is angiography. The flow chart also seems to advocate angiography after either renography or computed tomography regardless of the result of either. These investigations therefore do not seem to be contributory. Although renal vein thrombosis may be diagnosed on renal arterial angiography, renal venography is a more appropriate test. Ultrasound examination of the renal veins before any angiography would therefore seem advisable. Finally, the authors state that if a patient with acute loin pain is found to have a very small kidney no further investigation is warranted. We suggest that a very small kidney is an unlikely source of the patient's symptoms and that another cause should be sought. M J DUDDY S A BRADLEY S CHAPMAN
Department of Radiology, Children's Hospital, Birmingham B16 8ET 1 Stinchcombe SJ, Manhire AR, Bishop MC, Gregson RHS. Fibromuscular dysplasia of renal arteries: a neglected cause of acute loin pain. BrMedJ 1990;300:183-5. (20 January.)
Effective use of regional intensive therapy units SIR,-The paper by Dr Jane A M Purdie and colleagues' fails to answer the aims of the trial. Indeed the study design was such that it could not answer the objective of determining the effectiveness of regional intensive therapy units as results from regional units were not compared with those from units in district hospitals. I am assuming that the legend to the scatter diagram of APACHE score versus duration of admission before transfer is incorrect and that the symbols for survivors and deaths have been inadvertently switched. When all data on this diagram are analysed together by the MannWhitney U test then there is no significant difference between the survivors and those patients who died with regard to duration of illness before transfer. There is also no significant difference in APACHE score between the survivors and those who died. Moreover, of the patients who were transferred 10 days or more after the onset of their illness, 10 survived and nine died. These points refute the authors' suggestion that patients with an APACHE score greater than 10 who have been ill for more than 10 days should not be transferred to regional intensive therapy units. I support the concepts outlined in the document 469