far cry from implying that their use should be encouraged. MARK ASHTON MICK NIELSON DAVID SUTTON P'riicess Aninec Hospital,
Southampton S09 4HA \IcCabc M, Sibert JR, Rouitledge I'A. I'hosphate cncmas in childhood: a cause for concern. 1BMlI7 1991;302:1074. (4 Mma.2 2 Asltton MR. SuLtton D, Niclsen 1. Severe magnesitum toxicity aftcr magnesium sulphatc cicma in a chronically constipated chilid. 13,1,7 1990;300:541.
SIR,-In their lesson of the week on phosphate enemas in childhood' Dr M McCabe and colleagues correctly point out that the British datasheet for Fletchers' phosphate enema states: "Children: under 3, not recommended." In the past confusion may have been caused by the British National Formulary stating that the dose in children should be reduced according to age, but current editions are now consistent with the datasheet. Although absorption of phosphate has not been reported in the United Kingdom, the datasheet for the American counterpart (Fleet enema), which is more hypertonic (approximately 16% wt/vol sodium acid phosphate), records details of systemic
absorption.' Hypertonic phosphate enemas are commonly used in constipated subjects who do not respond to either dietary manipulation or oral treatment. The fact that over two million phosphate enemas are manufactured each year (Pharmax; data on file) confirms their continuing popularity. The main warning in the current British datasheet concerns the risk of inducing vasovagal episodes. We continue to emphasise the need for care in prescribing and administering phosphate enemas. Because they are almost always used in patients who are debilitated or elderly, or both, the possibility of untoward events is amplified, but such events may go unnoticed until too late. We have reviewed published reports on the safety of phosphate enemas. As well as problems caused by the constituents, injury caused by the enema nozzle occurs uncommonly but is probably underdiagnosed. In high risk patients-for example, patients with neurological impairment and those with communication difficulties -the rectum may unwittingly be perforated and chemical damage caused by the hypertonic phosphate may follow. (In one case reported in Britain rectal necrosis occurred and the patient required a defunctioning
colostomy.') With the British Data Sheet Compendium becoming available to the public,a just as the Physicians Desk Reference is on sale openly in the United States, patients may become as aware of the information therein as their doctors. Clearly, if products are used outside their strict guidelines questions may be expected. M GOLDMAN
IPharmax, Bcxle\, Kent DA5 INX I McCabe M, Sibert JR, Routlcdge PA. Phosphate enemas in childhood: cause for concern. BMI7 1991;302:1074. (4 Mai.) 2 Physicians desk reference. 45th ed. Oradell, New Jersey: Medical Economics, 1991. 3 Smith J, Carr N, Corrado OJ, Young A. Rectal necrosis after a phosphate enema. Age Ageing 1987;16:328-30. 4 Wells F. ABPI Xviewpoint. Association of Information Ojfice-rs in the Pharmameutical Indusir' Newsletter No 5 1990 October: 11-2.
Rectal examination in patients with abdominal pain SIR,-Dr Paul Kinnersley and colleagues made several comments' about our paper on rectal examination in patients with abdominal pain.2 They suggested that the sample of patients used in our study was a highly selected one. 1274
We stated in our paper that this was a consecutive series of patients seen at a district general hospital. Of the 1028 patients for whom we had data on rectal examination, 911 had been referred by their general practitioner, most of the remainder being self referred. The odds ratio, sensitivity, and specificity of rectal examination in the patients who had been referred by their general practitioner were almost identical with those in the patients who had referred themselves. This indicates that the profile of findings on rectal examination was independent of the source of the patient and suggests that the general practitioners' findings on rectal examination were not a discriminating factor for referral to hospital. Although this does suggest that rectal examination performed by general practitioners in patients with pain in the right lower quadrant might not be of value, we accept that on the basis of our data we cannot be certain that that is the case. Dr Kinnersley and colleagues also criticise the presentation of our data. They suggest that we should have given data on the specificity of signs for appendicitis. The specificity of a test conveys little useful information unless its sensitivity is also known because the diagnostic value of a test is related to the extent to which the sum of these two exceeds 100%. A test can have no diagnostic value but will have a high specificity if it occurs rarelyfor example, diagnosing only left handed patients as having appendicitis would give a specificity in the region of around 90%, but this would be offset by a very low sensitivity of around 10%. As the sum of the sensitivity and specificity does not exceed 100% this feature would clearly be of no diagnostic value. In our study the sensitivity of right sided rectal tenderness for acute appendicitis was 34% and the specificity 72%. The sum of these two figures only just exceeds 100%, indicating that right sided rectal tenderness, as we stated, is of little diagnostic value in diagnosing acute appendicitis. Comparing the figures for right sided rectal tenderness with those for rebound tenderness, which has a specificity of 62% and a sensitivity of 67%, shows that this particular test has substantial diagnostic potential. The odds ratio reflects both sensitivity and specificity and can be expressed as:
sensitivity x specificity (100 - sensitivity) x (100 - specificity) If the sum of the specificity and sensitivity is 100, this equation gives an odds ratio of 1. Clearly, odds ratios provide more useful information than is available from looking at specificity alone, particularly when the diagnostic values ofparticular signs are looked at. It is because of this that they are widely used as descriptive statistics in epidemiological studies, and they have a natural relation to analysis by multiple logistic regression. J MICHAEL DIXON R A ELTON University Department of Surgery, Royal Infirmary, Edinburgh EH3 9YW 1 Kinnersley P, Richards J, Owen P, Wilkinson C. Rectal examination in patients with abdominal pain. BMJ 1991;302: 908-9. (13 April.) 2 l)ixon JM, Elton RA, Rainey JB, Macleod DAD. Rectal examination in patients with pain in the right lower quadrant of the abdomen. BMJ, 1991;302:386-8. (16 February.)
Health education, cervical smears, and "Asian" women SIR,-Dr Brian R McAvoy and Rabia Raza's interesting study' was marred by their definition of the term "Asian"; this emphasises the need for accuracy and precision in definitions. They use the term Asian to refer to those "of New Commonwealth and Pakistani ethnic origin or descent, including those from Bangladesh and east Africa."
By their definition the study would include people of Chinese origin born in Singapore or Malaysia, and this is not what they intend. Perhaps they should have defined Asian as being those who are of Indian subcontinent origin or descent, including those from India, Pakistan, Bangladesh, and east Africa. One way of underlining the specific use of the term Asian is to put it in parentheses or italics, as some authors have done.2 HARPREET S KOHLI
Monklands and C umbernauld Unit, Lanarkshire Health Board, Coatbridge, Lanarkshire l\L5 3BN I McAvoy BR, Raza R. Can health education incrcase uptake of cerv'ical smear testing among Asian women? AM 7 1991;302:
833-6. (6 April.) 2 Kohli HS. The health of "Asians" in Britaini. Nattional Mfedical Journal of India 1988;1:27-30. 3 Firdous R, Bhopal RS. Rcproductive health of "Asiani" women: a comparative study with hospital and community perspcctives. Public Health 1999;103:307-15.
Benzodiazepines and pilot error SIR,-Ms Daphne Gloag's article on air crashes and human error highlighted the problems faced by air crews and traffic controllers.' A recent review has shown that, despite its dangers, the use of alcohol is still not uncommon among pilots.2 Time changes and jet lag may increase a pilot's urge to take a couple of drinks, and the related sleep disturbance is perhaps an understandable reason to consider the use of a hypnotic. To prevent residual sedation air crew and others whose occupations demand vigilance, motor skill, or decision making may be advised to take a short acting hypnotic such as triazolam, midazolam, or brotizolam.3 Because of triazolam's short half life of only two to three hours daytime sedation is rare, but amnesia-a recognised side effect of such short acting benzodiazepines-may persist after sedation has disappeared.4 This has become known as "traveller's amnesia. "9 As a result of using triazolam pilots may be suffering from impaired memory function without being aware of it. In other words, they may have done, or not done, things without having stored them in their memory. The potential danger of periodic amnesia in pilots has attracted little attention. In an authoritative review of the treatment of insomnia Gillin and Byerley briefly referred to the risk of traveller's amnesia but did not consider its hazard with regard to air traffic staff.6 As well as causing amnesia, drugs such as triazolam and midazolam have been associated with a variety of side effects, including increased daytime anxiety, behavioural abnormalities, and psychotic episodes.7'-" Selecting a safe hypnotic for pilots is difficult. It is not clear whether the ignition-interlock system, recently proposed as a convenient practical test for pilots,2 will be able to detect drug induced memory impairment. R H B MEYBOOM
'I'hc Netherlands Centre for Monitoring of Adverse Reactions to Drugs, PO Box 5406, 2280 HK Rijswijk, The Netherlands 1 Gloag D. Air crashes and human error. BMNJ 1991;302:550. (9 March.) 2 Model JG, Mountz JM. Drinking and flying-the problem of alcohol use by pilots. N Engli Med 1990;323:455-61. 3 Nicholson AN. Hypnotics and occupational medicine. fOccup Med 1990;32:335-41. 4 Scharf MB, Kauffman R, Brown L, Segal jj, Hirschowitz J. Morning amnestic effects of triazolam. Hillside J Clin Psychiatry 1986;8:38-45. S Morris HH, Estes 1ML. Traveler's armesia, transient global amnesia secondary to triazolam. JAMA 1987;258:945-6. 6 Gillin JC, Byerly WF. The diagnosis and management of insomnia. N Engli Med 1990;322:239-48. 7 Bixler EO, Kales A, Brubaker BH, Kales JD. Adverse reactions to benzodiazepine hypnotics: spontaneous reporting system. Phar7nacology 1987;35:286-300. 8 Tan TL, Bixler EO, Kales A, Cadieux RJ, Goodman AL.
BMJ
VOLUME 302
25 MAY 1991
Early morning insomnia, daytime anxietv, and organic mental disorder associated with triazolam. J Fam Pract
1985;20:592-4. 9 Regestein QR, Reich P. Agitation observed during treatment with newer hypnotic drugs. J Clin Psychiatrv 1985;46:280-3. 10 Patterson JF. Triazolam syndrome in the elderly. South Med]7 1987;80: 1425-6. 11 Fontaine R, Chouinard G. Annable L. Rebound anxiety in anxious patients after abrupt withdrawal of benzodiazepine trcatment. Amj Psvchiatrv 1984;141:848-52.
physiotherapy department by 36. The financial implications may not be minimal, as suggested by the authors, but in fact lead to a considerable increase in expenditure. F P MONSELL
M L PORTER Department of Orthopaedic Surgery, Royal Preston Hospital, Preston PR2 4HT I Packer GJ, Goring CG, Gayner AD, Craxford AD. Audit of ankle injuries in an accident and emnergencv department. BMJ 1991;302:885-7. (13 April.)
Audit of ankle injuries SIR,-Dr G J Packer and colleagues should be congratulated on their use of a protocol for managing ankle injuries in an accident and emergency department to improve treatment and reduce the cost of radiology, in terms of both expense and unnecessary irradiation of patients. Previous attempts have been made to apply clinical guidelines to the selection of patients for radiography in accident and emergency departments, most notably in the use of skull radiography after head injury; guidelines on this have been suggested by the Royal College of Radiologists.2 A recent study of the use of a modified version of these guidelines over 12 months suggested that it was easier to introduce than to sustain.' The rate of skull radiography initially fell by 40% but thereafter slowly increased so that by the end of the 12 months it had returned to the preimplementation value. The guidelines were introduced by displaying posters, distributing copies to casualty officers, and giving a short lecture to those starting a new job in the department. The authors comment that more attention should be given to sustaining any improvement resulting from using the guidelines rather than to further studies to justify their use. Using an algorithm attached to each relevant set of notes on initial presentation to the accident and emergency department' seems to be a logical step towards this. A review of the rate of completion of the protocol and the proportion of patients undergoing radiography 12 months after the algorithm was implemented would be valuable to assess this. In addition to using effective clinical guidelines for selective radiography, a significant and sustained reduction in the number of x ray examinations requested can be achieved when there is a concurrent teaching programme.4 Ideally, this should include regular lectures from radiologists, a film library of commonly missed lesions, and audit of x ray films.'
Cancer and HIV infection SIR, -In their recent editorial on cancer and HIV infection Drs Luke Hughes-Davies and Margaret Spittle state that they "know that the Centers for Disease Control classification may be missing some opportunistic infections."' Whether the surveillance definition for AIDS needs to be broadened or not, I am surprised to see our recently published work as the single reference to justify this statement.2 Our study documented bacteraemia in patients on admission to hospital in Nairobi, Kenya. Conventional pathogens, especially Salmonella typhimurium and Streptococcus pneumoniae, caused significantly more disease in seropositive than seronegative patients. In our discussion we developed the theme that many deaths related to HIV infection, particularly in poorer adults, are occurring relatively early in the course of HIV immunosuppression and are not caused by conditions used in the accepted diagnosis of AIDS. In the last sentence of the abstract we concluded that "the findings suggest non-opportunistic bacteria are important causes of morbidity and mortality in HIV-infected individuals in Africa." Nowhere did we state, or imply, that our seropositive patients had conditions that were being missed or should be reclassified as AIDS defining problems. Our paper is the first to highlight the importance of acute bacterial infections in patients positive for HIV antibody in Africa. We are delighted if it is quoted, but not if it is misread or misinterpreted. CHARLES GILKS
KemrilWellcome Trust Research Programme, Kenya Medical Research Institute, PO Box 43640, Nairobi, Kenya
N J A COZENS
Royal Infirmary of Edinburgh, Edinburgh EH3 9YW 1 Packer GJ, Goring CC, Gayner AD, Craxford AD. Audit of ankle injuries in an accident and emergency department. BMJ 1991;302:885-7. (13 April.) 2 Royal College of Radiologists. Costs and benefits of skull radiographv for head injury. Lancet 1981;ii:791-5. 3 Clarke JA, Adams JE. The application of clinical guidelines for skull radiograph' in the accident and emergency department: theory and practice. Clin Radtol 1990;41:152-5. 4 Gleadhill DNS, Thomsoni JY, Simms 1. Can more efficient use be made of x ray examinations in the accident and emergency department? BMJ7 1987;294:943-7. 5 Fielding JA. Improving accidettt and emergency radiology. Clin Radiol 1990;41:149-51.
SIR,-We disagree with some comments in the paper of Mr G J Packer and colleagues.' Although their study is to be commended for its reduction in radiography costs and inappropriate referrals to fracture clinic, we feel that it underplays the significance of the 18% increase in patients reviewed as a result. It is misleading to extrapolate savings in radiography to give annual figures while quoting weekly figures for patients reviewed in the accident department. If similar annual figures are calculated from data in the report, the number of patients reviewed in the accident department is increased by 216, in the soft tissue clinic by 60, and in the
BMJ
VOLUME
302
25
MAY
1991
1 Hughes-Davies L, Spittle M. Cancer and HIV infection. BM7 1991;302:673-4. (23 March.) 2 Gilks CF, Brindle RJ, Otieno LS, et al. Life threatening bacteraemia in HIV- I seropositive adults admitted to hospital in Nairobi, Kenya. Lancet 1990;336:545-9.
Biochemical screening for Down's syndrome SIR,-Professor Geoffrey Chamberlain notes that the cost of biochemical screening for Down's syndrome, when not set against savings in maintenance of such children, is used to delay introduction of this service,' and Dr T M Reynolds
suggests that a further reason has been uncertainty about which analytes to measure.2 Another reason may be the difficulty in providing adequate facilities for counselling patients. It has been usual to interpret the results of tests for (a fetoprotein as normal, raised, or borderline when screening for neural tube defects. In screening for Down's syndrome we report an actual risk and a suggested interpretation. Counselling is more complicated because interpretation depends on the perception of the stated risk by the patient and the obstetrician. In addition, knowledge of the effects of errors in estimation of gestational age' and of normally acceptable assay precision (table) may make the decision more difficult. Our contact with patients, general practitioners, and obstetricians suggests a need for study days on these topics in areas where biochemical screening for Down's syndrome is offered. It should be emphasised that despite these limitations biochemical screening is still likely to be three times more successful than age alone in identifying fetuses with Down's syndrome. These difficulties show, however, the need for a better discriminator for Down's syndrome and the importance of assay precision when the "triple test" is used. S HOLDING
Haematology Department, Kingston General Hospital, Kingston upon Hull HU3 I UR I Chamberlain G. ABC of antenatal care: detection and management of congenital abnormalities-II. BM7 1991;302:1013-6.
(27 April.) 2 Reynolds 1IM. Software for screening to assess risk of Down's
syndrome. BMJ 1991;302:965. (20 April.) 3 Holding S. Estimations of gestational age and screening for t)own's syndrome. BMJ 1991 ;302:%5. (20 April.)
Diagnosing acoustic neuroma SIR,-Drs I R C Swann and S Gatehouse are right to draw attention to the difficulties in the early diagnosis of acoustic neuroma.' The morbidity associated with surgical removal of tumours has repeatedly been shown to be directly related to tumour size. Major facial nerve complications, brain stem syndromes, and death are all more common with tumours of more than 2-5 cm diameter than with smaller lesions.2 The great majority of these patients present with audiovestibular symptoms (unilateral deafness, tinnitus, or imbalance) to the ear, nose, and throat clinic. Traditionally, considerable reliance has been placed on a battery of audiovestibular tests in the hope of achieving the ideal of early diagnosis by identifying those patients in whom it seemed reasonable to proceed to neuroradiological techniques of varying degrees of invasiveness (air or myodil meatography). These tests were based on the observation in retrocochlear pathology of excessive loudness adaption (tone decay), poor speech discrimination, and the absence of loudness recruitment (loudness balance test). More recently tests of stapedial reflex threshold and adaptation have been added, as has the auditory brainstem response. In the past decade the situation has changed. Firstly, the reliability of the traditional audiological
Effect of acceptable assay precision on likelihood ratio for Down's syndrome. Kits were used according to manufacturers' protocols Patient No 1 2 3 4 5
Age at estimated date of delivery (t Fetoprotein* Oestriol* (years) (kIU/1) (nmol/l) 390 30 5 289 30 4 37-6
55-4 93 9 268 53-9 81-6
4 20 8-58 253 7-72 3-37
Human chorionic gonadotrophin* (klU/I)
Risk of D)own's syndrome
Coefficienit of variation
45 3 19 7 940 12 8 112-1
1:735 1:19 544 1:110 1:13 047 1:157
31 1 24-6 478 21 4 395
of likelihood ratio (%/,)
*Mean assay coefficients of variation: (t fetoprotein 3-920/o (range 3 0-4-4); oestriol 8-581% (range 7-3-10-9); gonadotrophin 7 32% (range 5-3-9 8).
humatt chorionic 1275
Southampton S09 4HA \IcCabc M, Sibert JR, Rouitledge I'A. I'hosphate cncmas in childhood: a cause for concern. 1BMlI7 1991;302:1074. (4 Mma.2 2 Asltton MR. SuLtton D, Niclsen 1. Severe magnesitum toxicity aftcr magnesium sulphatc cicma in a chronically constipated chilid. 13,1,7 1990;300:541.
SIR,-In their lesson of the week on phosphate enemas in childhood' Dr M McCabe and colleagues correctly point out that the British datasheet for Fletchers' phosphate enema states: "Children: under 3, not recommended." In the past confusion may have been caused by the British National Formulary stating that the dose in children should be reduced according to age, but current editions are now consistent with the datasheet. Although absorption of phosphate has not been reported in the United Kingdom, the datasheet for the American counterpart (Fleet enema), which is more hypertonic (approximately 16% wt/vol sodium acid phosphate), records details of systemic
absorption.' Hypertonic phosphate enemas are commonly used in constipated subjects who do not respond to either dietary manipulation or oral treatment. The fact that over two million phosphate enemas are manufactured each year (Pharmax; data on file) confirms their continuing popularity. The main warning in the current British datasheet concerns the risk of inducing vasovagal episodes. We continue to emphasise the need for care in prescribing and administering phosphate enemas. Because they are almost always used in patients who are debilitated or elderly, or both, the possibility of untoward events is amplified, but such events may go unnoticed until too late. We have reviewed published reports on the safety of phosphate enemas. As well as problems caused by the constituents, injury caused by the enema nozzle occurs uncommonly but is probably underdiagnosed. In high risk patients-for example, patients with neurological impairment and those with communication difficulties -the rectum may unwittingly be perforated and chemical damage caused by the hypertonic phosphate may follow. (In one case reported in Britain rectal necrosis occurred and the patient required a defunctioning
colostomy.') With the British Data Sheet Compendium becoming available to the public,a just as the Physicians Desk Reference is on sale openly in the United States, patients may become as aware of the information therein as their doctors. Clearly, if products are used outside their strict guidelines questions may be expected. M GOLDMAN
IPharmax, Bcxle\, Kent DA5 INX I McCabe M, Sibert JR, Routlcdge PA. Phosphate enemas in childhood: cause for concern. BMI7 1991;302:1074. (4 Mai.) 2 Physicians desk reference. 45th ed. Oradell, New Jersey: Medical Economics, 1991. 3 Smith J, Carr N, Corrado OJ, Young A. Rectal necrosis after a phosphate enema. Age Ageing 1987;16:328-30. 4 Wells F. ABPI Xviewpoint. Association of Information Ojfice-rs in the Pharmameutical Indusir' Newsletter No 5 1990 October: 11-2.
Rectal examination in patients with abdominal pain SIR,-Dr Paul Kinnersley and colleagues made several comments' about our paper on rectal examination in patients with abdominal pain.2 They suggested that the sample of patients used in our study was a highly selected one. 1274
We stated in our paper that this was a consecutive series of patients seen at a district general hospital. Of the 1028 patients for whom we had data on rectal examination, 911 had been referred by their general practitioner, most of the remainder being self referred. The odds ratio, sensitivity, and specificity of rectal examination in the patients who had been referred by their general practitioner were almost identical with those in the patients who had referred themselves. This indicates that the profile of findings on rectal examination was independent of the source of the patient and suggests that the general practitioners' findings on rectal examination were not a discriminating factor for referral to hospital. Although this does suggest that rectal examination performed by general practitioners in patients with pain in the right lower quadrant might not be of value, we accept that on the basis of our data we cannot be certain that that is the case. Dr Kinnersley and colleagues also criticise the presentation of our data. They suggest that we should have given data on the specificity of signs for appendicitis. The specificity of a test conveys little useful information unless its sensitivity is also known because the diagnostic value of a test is related to the extent to which the sum of these two exceeds 100%. A test can have no diagnostic value but will have a high specificity if it occurs rarelyfor example, diagnosing only left handed patients as having appendicitis would give a specificity in the region of around 90%, but this would be offset by a very low sensitivity of around 10%. As the sum of the sensitivity and specificity does not exceed 100% this feature would clearly be of no diagnostic value. In our study the sensitivity of right sided rectal tenderness for acute appendicitis was 34% and the specificity 72%. The sum of these two figures only just exceeds 100%, indicating that right sided rectal tenderness, as we stated, is of little diagnostic value in diagnosing acute appendicitis. Comparing the figures for right sided rectal tenderness with those for rebound tenderness, which has a specificity of 62% and a sensitivity of 67%, shows that this particular test has substantial diagnostic potential. The odds ratio reflects both sensitivity and specificity and can be expressed as:
sensitivity x specificity (100 - sensitivity) x (100 - specificity) If the sum of the specificity and sensitivity is 100, this equation gives an odds ratio of 1. Clearly, odds ratios provide more useful information than is available from looking at specificity alone, particularly when the diagnostic values ofparticular signs are looked at. It is because of this that they are widely used as descriptive statistics in epidemiological studies, and they have a natural relation to analysis by multiple logistic regression. J MICHAEL DIXON R A ELTON University Department of Surgery, Royal Infirmary, Edinburgh EH3 9YW 1 Kinnersley P, Richards J, Owen P, Wilkinson C. Rectal examination in patients with abdominal pain. BMJ 1991;302: 908-9. (13 April.) 2 l)ixon JM, Elton RA, Rainey JB, Macleod DAD. Rectal examination in patients with pain in the right lower quadrant of the abdomen. BMJ, 1991;302:386-8. (16 February.)
Health education, cervical smears, and "Asian" women SIR,-Dr Brian R McAvoy and Rabia Raza's interesting study' was marred by their definition of the term "Asian"; this emphasises the need for accuracy and precision in definitions. They use the term Asian to refer to those "of New Commonwealth and Pakistani ethnic origin or descent, including those from Bangladesh and east Africa."
By their definition the study would include people of Chinese origin born in Singapore or Malaysia, and this is not what they intend. Perhaps they should have defined Asian as being those who are of Indian subcontinent origin or descent, including those from India, Pakistan, Bangladesh, and east Africa. One way of underlining the specific use of the term Asian is to put it in parentheses or italics, as some authors have done.2 HARPREET S KOHLI
Monklands and C umbernauld Unit, Lanarkshire Health Board, Coatbridge, Lanarkshire l\L5 3BN I McAvoy BR, Raza R. Can health education incrcase uptake of cerv'ical smear testing among Asian women? AM 7 1991;302:
833-6. (6 April.) 2 Kohli HS. The health of "Asians" in Britaini. Nattional Mfedical Journal of India 1988;1:27-30. 3 Firdous R, Bhopal RS. Rcproductive health of "Asiani" women: a comparative study with hospital and community perspcctives. Public Health 1999;103:307-15.
Benzodiazepines and pilot error SIR,-Ms Daphne Gloag's article on air crashes and human error highlighted the problems faced by air crews and traffic controllers.' A recent review has shown that, despite its dangers, the use of alcohol is still not uncommon among pilots.2 Time changes and jet lag may increase a pilot's urge to take a couple of drinks, and the related sleep disturbance is perhaps an understandable reason to consider the use of a hypnotic. To prevent residual sedation air crew and others whose occupations demand vigilance, motor skill, or decision making may be advised to take a short acting hypnotic such as triazolam, midazolam, or brotizolam.3 Because of triazolam's short half life of only two to three hours daytime sedation is rare, but amnesia-a recognised side effect of such short acting benzodiazepines-may persist after sedation has disappeared.4 This has become known as "traveller's amnesia. "9 As a result of using triazolam pilots may be suffering from impaired memory function without being aware of it. In other words, they may have done, or not done, things without having stored them in their memory. The potential danger of periodic amnesia in pilots has attracted little attention. In an authoritative review of the treatment of insomnia Gillin and Byerley briefly referred to the risk of traveller's amnesia but did not consider its hazard with regard to air traffic staff.6 As well as causing amnesia, drugs such as triazolam and midazolam have been associated with a variety of side effects, including increased daytime anxiety, behavioural abnormalities, and psychotic episodes.7'-" Selecting a safe hypnotic for pilots is difficult. It is not clear whether the ignition-interlock system, recently proposed as a convenient practical test for pilots,2 will be able to detect drug induced memory impairment. R H B MEYBOOM
'I'hc Netherlands Centre for Monitoring of Adverse Reactions to Drugs, PO Box 5406, 2280 HK Rijswijk, The Netherlands 1 Gloag D. Air crashes and human error. BMNJ 1991;302:550. (9 March.) 2 Model JG, Mountz JM. Drinking and flying-the problem of alcohol use by pilots. N Engli Med 1990;323:455-61. 3 Nicholson AN. Hypnotics and occupational medicine. fOccup Med 1990;32:335-41. 4 Scharf MB, Kauffman R, Brown L, Segal jj, Hirschowitz J. Morning amnestic effects of triazolam. Hillside J Clin Psychiatry 1986;8:38-45. S Morris HH, Estes 1ML. Traveler's armesia, transient global amnesia secondary to triazolam. JAMA 1987;258:945-6. 6 Gillin JC, Byerly WF. The diagnosis and management of insomnia. N Engli Med 1990;322:239-48. 7 Bixler EO, Kales A, Brubaker BH, Kales JD. Adverse reactions to benzodiazepine hypnotics: spontaneous reporting system. Phar7nacology 1987;35:286-300. 8 Tan TL, Bixler EO, Kales A, Cadieux RJ, Goodman AL.
BMJ
VOLUME 302
25 MAY 1991
Early morning insomnia, daytime anxietv, and organic mental disorder associated with triazolam. J Fam Pract
1985;20:592-4. 9 Regestein QR, Reich P. Agitation observed during treatment with newer hypnotic drugs. J Clin Psychiatrv 1985;46:280-3. 10 Patterson JF. Triazolam syndrome in the elderly. South Med]7 1987;80: 1425-6. 11 Fontaine R, Chouinard G. Annable L. Rebound anxiety in anxious patients after abrupt withdrawal of benzodiazepine trcatment. Amj Psvchiatrv 1984;141:848-52.
physiotherapy department by 36. The financial implications may not be minimal, as suggested by the authors, but in fact lead to a considerable increase in expenditure. F P MONSELL
M L PORTER Department of Orthopaedic Surgery, Royal Preston Hospital, Preston PR2 4HT I Packer GJ, Goring CG, Gayner AD, Craxford AD. Audit of ankle injuries in an accident and emnergencv department. BMJ 1991;302:885-7. (13 April.)
Audit of ankle injuries SIR,-Dr G J Packer and colleagues should be congratulated on their use of a protocol for managing ankle injuries in an accident and emergency department to improve treatment and reduce the cost of radiology, in terms of both expense and unnecessary irradiation of patients. Previous attempts have been made to apply clinical guidelines to the selection of patients for radiography in accident and emergency departments, most notably in the use of skull radiography after head injury; guidelines on this have been suggested by the Royal College of Radiologists.2 A recent study of the use of a modified version of these guidelines over 12 months suggested that it was easier to introduce than to sustain.' The rate of skull radiography initially fell by 40% but thereafter slowly increased so that by the end of the 12 months it had returned to the preimplementation value. The guidelines were introduced by displaying posters, distributing copies to casualty officers, and giving a short lecture to those starting a new job in the department. The authors comment that more attention should be given to sustaining any improvement resulting from using the guidelines rather than to further studies to justify their use. Using an algorithm attached to each relevant set of notes on initial presentation to the accident and emergency department' seems to be a logical step towards this. A review of the rate of completion of the protocol and the proportion of patients undergoing radiography 12 months after the algorithm was implemented would be valuable to assess this. In addition to using effective clinical guidelines for selective radiography, a significant and sustained reduction in the number of x ray examinations requested can be achieved when there is a concurrent teaching programme.4 Ideally, this should include regular lectures from radiologists, a film library of commonly missed lesions, and audit of x ray films.'
Cancer and HIV infection SIR, -In their recent editorial on cancer and HIV infection Drs Luke Hughes-Davies and Margaret Spittle state that they "know that the Centers for Disease Control classification may be missing some opportunistic infections."' Whether the surveillance definition for AIDS needs to be broadened or not, I am surprised to see our recently published work as the single reference to justify this statement.2 Our study documented bacteraemia in patients on admission to hospital in Nairobi, Kenya. Conventional pathogens, especially Salmonella typhimurium and Streptococcus pneumoniae, caused significantly more disease in seropositive than seronegative patients. In our discussion we developed the theme that many deaths related to HIV infection, particularly in poorer adults, are occurring relatively early in the course of HIV immunosuppression and are not caused by conditions used in the accepted diagnosis of AIDS. In the last sentence of the abstract we concluded that "the findings suggest non-opportunistic bacteria are important causes of morbidity and mortality in HIV-infected individuals in Africa." Nowhere did we state, or imply, that our seropositive patients had conditions that were being missed or should be reclassified as AIDS defining problems. Our paper is the first to highlight the importance of acute bacterial infections in patients positive for HIV antibody in Africa. We are delighted if it is quoted, but not if it is misread or misinterpreted. CHARLES GILKS
KemrilWellcome Trust Research Programme, Kenya Medical Research Institute, PO Box 43640, Nairobi, Kenya
N J A COZENS
Royal Infirmary of Edinburgh, Edinburgh EH3 9YW 1 Packer GJ, Goring CC, Gayner AD, Craxford AD. Audit of ankle injuries in an accident and emergency department. BMJ 1991;302:885-7. (13 April.) 2 Royal College of Radiologists. Costs and benefits of skull radiographv for head injury. Lancet 1981;ii:791-5. 3 Clarke JA, Adams JE. The application of clinical guidelines for skull radiograph' in the accident and emergency department: theory and practice. Clin Radtol 1990;41:152-5. 4 Gleadhill DNS, Thomsoni JY, Simms 1. Can more efficient use be made of x ray examinations in the accident and emergency department? BMJ7 1987;294:943-7. 5 Fielding JA. Improving accidettt and emergency radiology. Clin Radiol 1990;41:149-51.
SIR,-We disagree with some comments in the paper of Mr G J Packer and colleagues.' Although their study is to be commended for its reduction in radiography costs and inappropriate referrals to fracture clinic, we feel that it underplays the significance of the 18% increase in patients reviewed as a result. It is misleading to extrapolate savings in radiography to give annual figures while quoting weekly figures for patients reviewed in the accident department. If similar annual figures are calculated from data in the report, the number of patients reviewed in the accident department is increased by 216, in the soft tissue clinic by 60, and in the
BMJ
VOLUME
302
25
MAY
1991
1 Hughes-Davies L, Spittle M. Cancer and HIV infection. BM7 1991;302:673-4. (23 March.) 2 Gilks CF, Brindle RJ, Otieno LS, et al. Life threatening bacteraemia in HIV- I seropositive adults admitted to hospital in Nairobi, Kenya. Lancet 1990;336:545-9.
Biochemical screening for Down's syndrome SIR,-Professor Geoffrey Chamberlain notes that the cost of biochemical screening for Down's syndrome, when not set against savings in maintenance of such children, is used to delay introduction of this service,' and Dr T M Reynolds
suggests that a further reason has been uncertainty about which analytes to measure.2 Another reason may be the difficulty in providing adequate facilities for counselling patients. It has been usual to interpret the results of tests for (a fetoprotein as normal, raised, or borderline when screening for neural tube defects. In screening for Down's syndrome we report an actual risk and a suggested interpretation. Counselling is more complicated because interpretation depends on the perception of the stated risk by the patient and the obstetrician. In addition, knowledge of the effects of errors in estimation of gestational age' and of normally acceptable assay precision (table) may make the decision more difficult. Our contact with patients, general practitioners, and obstetricians suggests a need for study days on these topics in areas where biochemical screening for Down's syndrome is offered. It should be emphasised that despite these limitations biochemical screening is still likely to be three times more successful than age alone in identifying fetuses with Down's syndrome. These difficulties show, however, the need for a better discriminator for Down's syndrome and the importance of assay precision when the "triple test" is used. S HOLDING
Haematology Department, Kingston General Hospital, Kingston upon Hull HU3 I UR I Chamberlain G. ABC of antenatal care: detection and management of congenital abnormalities-II. BM7 1991;302:1013-6.
(27 April.) 2 Reynolds 1IM. Software for screening to assess risk of Down's
syndrome. BMJ 1991;302:965. (20 April.) 3 Holding S. Estimations of gestational age and screening for t)own's syndrome. BMJ 1991 ;302:%5. (20 April.)
Diagnosing acoustic neuroma SIR,-Drs I R C Swann and S Gatehouse are right to draw attention to the difficulties in the early diagnosis of acoustic neuroma.' The morbidity associated with surgical removal of tumours has repeatedly been shown to be directly related to tumour size. Major facial nerve complications, brain stem syndromes, and death are all more common with tumours of more than 2-5 cm diameter than with smaller lesions.2 The great majority of these patients present with audiovestibular symptoms (unilateral deafness, tinnitus, or imbalance) to the ear, nose, and throat clinic. Traditionally, considerable reliance has been placed on a battery of audiovestibular tests in the hope of achieving the ideal of early diagnosis by identifying those patients in whom it seemed reasonable to proceed to neuroradiological techniques of varying degrees of invasiveness (air or myodil meatography). These tests were based on the observation in retrocochlear pathology of excessive loudness adaption (tone decay), poor speech discrimination, and the absence of loudness recruitment (loudness balance test). More recently tests of stapedial reflex threshold and adaptation have been added, as has the auditory brainstem response. In the past decade the situation has changed. Firstly, the reliability of the traditional audiological
Effect of acceptable assay precision on likelihood ratio for Down's syndrome. Kits were used according to manufacturers' protocols Patient No 1 2 3 4 5
Age at estimated date of delivery (t Fetoprotein* Oestriol* (years) (kIU/1) (nmol/l) 390 30 5 289 30 4 37-6
55-4 93 9 268 53-9 81-6
4 20 8-58 253 7-72 3-37
Human chorionic gonadotrophin* (klU/I)
Risk of D)own's syndrome
Coefficienit of variation
45 3 19 7 940 12 8 112-1
1:735 1:19 544 1:110 1:13 047 1:157
31 1 24-6 478 21 4 395
of likelihood ratio (%/,)
*Mean assay coefficients of variation: (t fetoprotein 3-920/o (range 3 0-4-4); oestriol 8-581% (range 7-3-10-9); gonadotrophin 7 32% (range 5-3-9 8).
humatt chorionic 1275
