high scores and little room for improvement (that is, there was a ceiling effect). The section on evaluation also mentions the use of several other tests, including a letter cancellation task, a 10 word recognition task, and paired associate memory tests. The results of tests are omitted from the results section. Thus the methodology of this trial makes it impossible to draw any conclusions about the effect of tacrine, and the negative results reported are as open to doubt as the positive results in the original paper by Summers et al,< which Dr Chatellier and colleagues have attempted to replicate.
population. Thus the recruitment was biased towards some subgroups-that is, those with familial forms of the disease, as underlined by Professor Levy. Also, Alzheimer's disease is not a homogeneous disease, as pointed out by Dr Wilcock, and some subgroups may be sensitive to tacrine. Our study does not make it possible to address these two points. We hope that this additional information will be provided by the large studies going on in Canada and United States. Presently, we can conclude only that the benefit to risk ratio of tacrine is such that its use cannot be recommended in all patients with Alzheimer's disease. G CHATELLIER
RAYMOND LEVY Institute of Psychiatry, London SE5 8AF 1 Chatellier G, Lacomblez Z on behalf of Groupe Francais d'Etuide de la Fetrahvdroaminoacridine. TIacrine tetrahydroaminoacridine; THA) and lecithin in senile dementia of the Alzheimer type: a multicentre trial. frM _dJ 1990;300:495-9. (24 February.) 2 Little A, Lesy R, Chuaqui-Kidd P', Hand D. A double-blind placebo-controlled trial of high-dose lecithin in Alzheimer's disease. 7 Neurol Neurosurg Pssschtatrv 1985;48:736-42. 3 Summers WK, Majowski LV, Marsh GMI, Tachiki K, Kling A. Oral tetrahydroaminoacridine in long-term treatment of senile dementia. NEnggl7Med 1986;315:1241-5.
AUTHOR'S REPLY,-There are no standardised methods for neuropharmacological trials in patients with Alzheimer's disease. Consequently, we chose to replicate the study of Summers et al' in some of its aspects as this study triggered the present worldwide research on tacrine (tetrahydroaminoacridine; THA). We agree with most comments with respect to the absence of a washout period and pointed out this problem in our discussion. When we designed our study data on the pharmacology of tacrine were still very scarce, and there is now some evidence that the drug has a carryover effect for some weeks, as underlined by Dr G K Wilcock. We did not observe any carryover effect as we did not observe any effect. This remains true when we compare the results obtained during the first period of treatment between the two groups (table III), but the power of the study becomes much smaller. Period effects are one of the major problems of crossover studies and are not always avoided by long washout periods, as suggested by Drs J R Murray and C B Blakemore. For example, during a six month crossover study evolution of the disease may puzzle evaluation of the drug efficacy. We agree with Professor Raymond Levy that the mini mental state and the Stockton scales may not have been completely appropriate to measure change in our patients, but no change was detected by the other cognitive instruments we used (letter cancellation task, 10 word recognition task, and paired associate memory tests). Professor Levy also points out that side effects could have unblinded the investigator. This may explain the small improvement scored by the doctor on the visual analogue scale, contrasting with the absence of alteration on the other scales. Anyway, unblinding usually favours the active drug. To conclude on the results of the first part of our study we would like to emphasise that, with a comparable methodology, we did not observe the very encouraging results obtained by Summers et al after short term administration of tacrine. With regard to the long term effects of the drug, each patient or patient's family, or both, had the opportunity to choose the best treatment (tacrine or placebo) at the end of the crossover and to continue the trial for four additional months. Again no obvious benefit was observed in the 30 patients who were given tacrine for four months, although we had expected that those patients would be the most responsive to the drug. Our conclusions may rely strongly on the population studied. All patients were recruited in specialised centres and not from the general
940
Service d'Informatique Mddicale, CHU Piti&-SalpWtricre, 75634 Paris, France I Summers WK, MIajowski LV, Marsh GM, Tachiki K, Kling A. Oral tetrahydroaminoacridine in long-term treatment of senile dementia. N Engli Med 1986;315:1241-5.
SIR,-The recent publication by Dr G Chatellier and colleagues on the treatment of patients with senile dementia of the Alzheimer type' may cause some doctors to conclude prematurely that cholinesterase inhibitors have no place in such treatment. A major problem in any trial of treatment in patients with Alzheimer's disease is to establish the diagnosis. We note that Dr Chatellier and colleagues pursued all of the routine selection criteria for their patients but included up to 10% of patients with a high risk of atherosclerotic brain disease. These patients may not respond in the same manner as those with Alzheimer's disease. Even among carefully selected patients there may still be differences that could lead to identifying groups of "responders and non-responders."2 Though we agree that tacrine (tetrahydroaminoacridine; THA) is unlikely to be used widely because of its toxicity, it nevertheless is the most promising drug to date. If it has any effect it couild act as a core compound from which to develop others. P C WOOD C M CASTLEDEN
Department of Medicine for the Elderly, Leicester General Hospital, Leicester LE5 4PW I Chatellier G, Lacomblez L on behalfofGroupe Franais d'Etude de la Tetrahydroaminoacridine. Tacrine (tetrahydroaminoacridine; THA) and lecithin in senile dementia of the Alzheimer type: a multicentre trial. BrMedj 1990;300:495-9.
(24 February.) 2 Harrell LE, Jope RS, Falgout J, et al. Biological and neuropsychological characterisation of physostigmine and nonresponders in Alzheimer's disease. J Am Geriatr Soc 1990;38:
113-22.
SIR,-Although the collaborative study of tacrine (tetrahydroaminoacridine; THA) is the largest yet published,' it is difficult to draw from it inferences about the drug's efficacy in patients with Alzheimer's dementia. The authors' data do not support their conclusion that treatment with tacrine and lecithin does not improve the cognitive and behavioural symptoms of patients with Alzheimer's disease. At best the authors might conclude that their results did not show tacrine to be significantly superior to placebo but that the power of their study was low. A randomised parallel group study with a longer treatment period and larger sample would have been far more appropriate than a crossover study in a population with dementia in which deterioration and a large amount of interindividual variation is expected. The authors may have found a crossover study to be more attractive because of its apparent requirement for fewer patients. But their analysis is confounded by the lack of a washout, rapid dosage titration to 125 mg/day or until side effects
occurred, and allowing over 35%, of the sample to receive concurrent benzodiazepines (for insomnia), haloperidol (for psychosis), and domperidone (for nausea or vomiting). The rapid titration probably resulted in 24 of 60 patients having had moderate gastrointestinal side effects; an additional six had asymptomatic hepatitis. All of this scrves to compromise the masking of treatment and the observations of the effect of treatment. The authors acknowledge the possibility of carryover effects but dismiss their significance by arguing that as no effects were detected neurologically (two patients developed hepatitis as a carryover effect) there was little likelihood of therapeutic carryover effects. They attempted to compensate for their short (four week) treatmcnt trial by allowing a non-random assignment bv relatives' blinded choice to placebo or tacrine for an additional four months. Treatment trials with nonrandom assignment are also difficult to interpret. The data presented are interesting in that tacrine consistently showed a greater effect than placebo. In order to estimate its magnitude without the confounding of the crossover design the effects on mini mental state examination scores (adjusted for the within group pretreatment versus posttreatment scores) were calculated2 for the two treatment groups based on the randomly assigned treatment received during the first 28 days (table III). The overall size of the effect was d=0 28, a small to moderate effect, where d is the ratio of the difference between treatment means to the pooled standard deviation. Effect sizes estimated from thc relatives' and doctors' visual analogue scale rating of improvement were approximately d--023 (relatives) and d=0 48 (doctors). Lastly, relatives chose tacrine for extended treatment 58% of the time compared with placebo 42% (approximate d= 0-32). Although these effect sizes vary, their average (d=0 33) is approximately the effect observed in a previous uncontrolled pilot study of 46 patients with Alzheimer's disease who received tacrine and lecithin for eight weeks. In order to have an 80'S, chance to detect a significant difference at the (t=0 05 level a parallel group placebo controlled trial would require 148 subjects in each treatment group (or 116 with a one tailed test).' Despite the limitations of its design the study provides evidence of a modest clinical effect for tacrine in patients with Alzheimer's disease. Adequately large and well designed multicentred studies are needed to enable reasonable decisions to be made about tacrine's clinical significance and potential utility. In the interim there will be "inconclusive, contradictory trials of small size from which it is almost impossible to make generalisations for clinical practice. "4 LON S SCHNEIDER
Department of Psychiatry, University of Southern California, Los Angeles, California 90033, United States I Chatellier G, Lacomblez L on behalf ot(iroupe liraintais d'I(tude de la Tetrahydroaminoacridinie. racrine (tetrahvdroaminoacridine; rHA) and lecithin in senile dementia of the Alzheimer type: a multicentrc trial. BreIed] 1990;300:495-9.
(24 February.) 2 Cohen J. Statistical ptrwer analvsis Jor the behavioral sciences. Hillsdale, New Jersey: Lawrence Erlbaum Associates, 1988. 3 Gauthier S, Bouchard R, Bacher Y, et al. Progress ots the Canadian multicenter trial of tetrahydroanminoacridinc ssith
lecithin in Alzhcimer's disease. Can 16: 543-6. 4
Goodmats NW. Multicetntre trials.
Br
J7 \eueol
ScI 1989;
tIed] 1990;300:538. (24
Fcbruiary.)
Pocket programmable computers in clinical practice SIR,-Dr Jeremy Wyatt commented on the potential value of putting clinical scoring systems on a pocket sized programmable computer.' We have
BMJ VOLUME 300
7 APRIL 1990
been developing such systems for the management of a varietv of medical conditions, of which a system for the diagnosis, physical examination, and treatment of hvpercholesterolaemia is currentlv undergoing field trials with general practitioners in the west midlands. The system is supplied for the use of a Psion Organiser and incorporates regional knowledge in a particular subject as appropriate for use in primary care or by junior hospital medical staff. The medical knowledge supplied facilitates patient management up to a level considered appropriate for the general practitioner or non-specialist physician and contains locally determined referral criteria for specialist attention. Preliminary results suggest that such a format is acceptable to both general practitioners and practice nurses and that programmable pocket computers are an appropriate vehicle for the devolution of medical expertise to nonspecialists or non-medical staff such as practice nurses. Inappropriate referrals to hospital clinics mav also be reduced by providing clear guidelines to referring general practitioners with this medium. We also envisage using such systems for providing useful advice on foreign travel prophylaxis, poisoning, laboratory sampling requirements, protocols for standard medical tests, etc. The potential for devolution of medical expertise is just beginning to be realised. Pocket sized computers can retain and instantaneously recall a large amount of medically relevant data and, unlike a text book, can be readily updated, are portable, and have a variety of other functions. M F RYAN
N CORBETT
I CLARK
M PETERS
Department of ledical Computing, Wolfson Research Laboratories, Queen Elizabeth Medical Centre, Birtningham B15 2FH 1 Wyatt J. Construction of clinical scoring systems. Br Med J
1990;300:538-9. (24 February-.'l
MIF, Clark I, P'etcrs .Ml, Cramb R. A low cost indicator for rationalisation of lipid clinic refcrrals. In: Halloran SP, ed. Focus '89. Londot: Association of Clinical Biochemists,
2 Ryan
1989:68.
Streptokinase in acute aortic dissection SIR,-The report by Mr J Butler and colleagues emphasises the risk of leaving major medical management decisions to inexperienced junior medical staff.' In each of the cases described the house officer was apparently the first medical practitioner to see patients admitted as emergencies. This is an illogical method for assessing patients. The administration of potentially life saving but also potentially life threatening treatments surely requires a higher level of skill and experience than should be expected from a house officer. It is noteworthy that in all of the patients described in this report aortic dissection should have been considered on the basis of the chest radiograph, which was available to the admitting house officer. There is insufficient data in the report by Dr Nicholas P Curzen and colleagues2 to judge whether the same circumstances occurred in their patients, although in one case the chest radiograph was suggestive of aortic dissection. In the study by Mr Butler and colleagues it is not surprising that the diagnosis of aortic dissection was not considered by the admitting house officer, both in view of the limited experience in cardiovascular diseases that most medical students receive during their training and also the limited exposure to training in radiology in most undergraduate medical training. The problems concerned in distinguishing acute aortic dissection from other causes of severe acute chest pain should not cause the uses of thrombolytic agents to be reduced. Rather it is
BMJ
VOLUME
300
7
APRIL
1990
an important reason for changing the way in which patients admitted with acute conditions are assessed. Surely it is no longer satisfactory to delegate assessment of acutely ill patients to the most junior members of the medical team; this should now be the responsibility of a more senior (that is, registrar grade) admitting physician. This will improve the accuracy of the initial management of these patients and will help to prevent the potential catastrophes described by Mr Butler and colleagues. It may also help to prevent an epidemic of unnecessary requests for investigation for possible aortic dissection before doctors administer thrombolytic treatment. G HARTNELL
Department of Radiodiagnosis, Bristol Royal Infirmary, Bristol BS2 8HW I Butler J, Davies AH, Wecstab- S. Streptokinase in acute aortic dissection. BrMteedJ 1990;300:517-9. (24 February.) 2 Curzen NP, Clarke B, Gray HH. Intravenous thrombolysis for suspected myocardial infarction: a cautionary note. Br MedJ7 1990;300:513. (24 February.)
Medical problems of adults who were sexually abused in childhood SIR,-Assessing the possible effects of adversity during childhood on health in adulthood requires studies that either include appropriate control subjects for comparison or eliminate the effects of confounding variables statistically. In their study on the medical problems of adults who were sexually abused in childhood Dr R P Arnold and colleagues used neither of these techniques.' This leads to most basic problems in interpreting their results. For example, we would surely expect the total number of medical specialties contacted by any person (whether sexually abused in childhood or not) to increase with age. The data presented (tables I and III) show exactly this expected correlation between age and the number of specialties contacted for the seven abused women (Spearman's test, Q=088; z=2 16; p
population. Thus the recruitment was biased towards some subgroups-that is, those with familial forms of the disease, as underlined by Professor Levy. Also, Alzheimer's disease is not a homogeneous disease, as pointed out by Dr Wilcock, and some subgroups may be sensitive to tacrine. Our study does not make it possible to address these two points. We hope that this additional information will be provided by the large studies going on in Canada and United States. Presently, we can conclude only that the benefit to risk ratio of tacrine is such that its use cannot be recommended in all patients with Alzheimer's disease. G CHATELLIER
RAYMOND LEVY Institute of Psychiatry, London SE5 8AF 1 Chatellier G, Lacomblez Z on behalf of Groupe Francais d'Etuide de la Fetrahvdroaminoacridine. TIacrine tetrahydroaminoacridine; THA) and lecithin in senile dementia of the Alzheimer type: a multicentre trial. frM _dJ 1990;300:495-9. (24 February.) 2 Little A, Lesy R, Chuaqui-Kidd P', Hand D. A double-blind placebo-controlled trial of high-dose lecithin in Alzheimer's disease. 7 Neurol Neurosurg Pssschtatrv 1985;48:736-42. 3 Summers WK, Majowski LV, Marsh GMI, Tachiki K, Kling A. Oral tetrahydroaminoacridine in long-term treatment of senile dementia. NEnggl7Med 1986;315:1241-5.
AUTHOR'S REPLY,-There are no standardised methods for neuropharmacological trials in patients with Alzheimer's disease. Consequently, we chose to replicate the study of Summers et al' in some of its aspects as this study triggered the present worldwide research on tacrine (tetrahydroaminoacridine; THA). We agree with most comments with respect to the absence of a washout period and pointed out this problem in our discussion. When we designed our study data on the pharmacology of tacrine were still very scarce, and there is now some evidence that the drug has a carryover effect for some weeks, as underlined by Dr G K Wilcock. We did not observe any carryover effect as we did not observe any effect. This remains true when we compare the results obtained during the first period of treatment between the two groups (table III), but the power of the study becomes much smaller. Period effects are one of the major problems of crossover studies and are not always avoided by long washout periods, as suggested by Drs J R Murray and C B Blakemore. For example, during a six month crossover study evolution of the disease may puzzle evaluation of the drug efficacy. We agree with Professor Raymond Levy that the mini mental state and the Stockton scales may not have been completely appropriate to measure change in our patients, but no change was detected by the other cognitive instruments we used (letter cancellation task, 10 word recognition task, and paired associate memory tests). Professor Levy also points out that side effects could have unblinded the investigator. This may explain the small improvement scored by the doctor on the visual analogue scale, contrasting with the absence of alteration on the other scales. Anyway, unblinding usually favours the active drug. To conclude on the results of the first part of our study we would like to emphasise that, with a comparable methodology, we did not observe the very encouraging results obtained by Summers et al after short term administration of tacrine. With regard to the long term effects of the drug, each patient or patient's family, or both, had the opportunity to choose the best treatment (tacrine or placebo) at the end of the crossover and to continue the trial for four additional months. Again no obvious benefit was observed in the 30 patients who were given tacrine for four months, although we had expected that those patients would be the most responsive to the drug. Our conclusions may rely strongly on the population studied. All patients were recruited in specialised centres and not from the general
940
Service d'Informatique Mddicale, CHU Piti&-SalpWtricre, 75634 Paris, France I Summers WK, MIajowski LV, Marsh GM, Tachiki K, Kling A. Oral tetrahydroaminoacridine in long-term treatment of senile dementia. N Engli Med 1986;315:1241-5.
SIR,-The recent publication by Dr G Chatellier and colleagues on the treatment of patients with senile dementia of the Alzheimer type' may cause some doctors to conclude prematurely that cholinesterase inhibitors have no place in such treatment. A major problem in any trial of treatment in patients with Alzheimer's disease is to establish the diagnosis. We note that Dr Chatellier and colleagues pursued all of the routine selection criteria for their patients but included up to 10% of patients with a high risk of atherosclerotic brain disease. These patients may not respond in the same manner as those with Alzheimer's disease. Even among carefully selected patients there may still be differences that could lead to identifying groups of "responders and non-responders."2 Though we agree that tacrine (tetrahydroaminoacridine; THA) is unlikely to be used widely because of its toxicity, it nevertheless is the most promising drug to date. If it has any effect it couild act as a core compound from which to develop others. P C WOOD C M CASTLEDEN
Department of Medicine for the Elderly, Leicester General Hospital, Leicester LE5 4PW I Chatellier G, Lacomblez L on behalfofGroupe Franais d'Etude de la Tetrahydroaminoacridine. Tacrine (tetrahydroaminoacridine; THA) and lecithin in senile dementia of the Alzheimer type: a multicentre trial. BrMedj 1990;300:495-9.
(24 February.) 2 Harrell LE, Jope RS, Falgout J, et al. Biological and neuropsychological characterisation of physostigmine and nonresponders in Alzheimer's disease. J Am Geriatr Soc 1990;38:
113-22.
SIR,-Although the collaborative study of tacrine (tetrahydroaminoacridine; THA) is the largest yet published,' it is difficult to draw from it inferences about the drug's efficacy in patients with Alzheimer's dementia. The authors' data do not support their conclusion that treatment with tacrine and lecithin does not improve the cognitive and behavioural symptoms of patients with Alzheimer's disease. At best the authors might conclude that their results did not show tacrine to be significantly superior to placebo but that the power of their study was low. A randomised parallel group study with a longer treatment period and larger sample would have been far more appropriate than a crossover study in a population with dementia in which deterioration and a large amount of interindividual variation is expected. The authors may have found a crossover study to be more attractive because of its apparent requirement for fewer patients. But their analysis is confounded by the lack of a washout, rapid dosage titration to 125 mg/day or until side effects
occurred, and allowing over 35%, of the sample to receive concurrent benzodiazepines (for insomnia), haloperidol (for psychosis), and domperidone (for nausea or vomiting). The rapid titration probably resulted in 24 of 60 patients having had moderate gastrointestinal side effects; an additional six had asymptomatic hepatitis. All of this scrves to compromise the masking of treatment and the observations of the effect of treatment. The authors acknowledge the possibility of carryover effects but dismiss their significance by arguing that as no effects were detected neurologically (two patients developed hepatitis as a carryover effect) there was little likelihood of therapeutic carryover effects. They attempted to compensate for their short (four week) treatmcnt trial by allowing a non-random assignment bv relatives' blinded choice to placebo or tacrine for an additional four months. Treatment trials with nonrandom assignment are also difficult to interpret. The data presented are interesting in that tacrine consistently showed a greater effect than placebo. In order to estimate its magnitude without the confounding of the crossover design the effects on mini mental state examination scores (adjusted for the within group pretreatment versus posttreatment scores) were calculated2 for the two treatment groups based on the randomly assigned treatment received during the first 28 days (table III). The overall size of the effect was d=0 28, a small to moderate effect, where d is the ratio of the difference between treatment means to the pooled standard deviation. Effect sizes estimated from thc relatives' and doctors' visual analogue scale rating of improvement were approximately d--023 (relatives) and d=0 48 (doctors). Lastly, relatives chose tacrine for extended treatment 58% of the time compared with placebo 42% (approximate d= 0-32). Although these effect sizes vary, their average (d=0 33) is approximately the effect observed in a previous uncontrolled pilot study of 46 patients with Alzheimer's disease who received tacrine and lecithin for eight weeks. In order to have an 80'S, chance to detect a significant difference at the (t=0 05 level a parallel group placebo controlled trial would require 148 subjects in each treatment group (or 116 with a one tailed test).' Despite the limitations of its design the study provides evidence of a modest clinical effect for tacrine in patients with Alzheimer's disease. Adequately large and well designed multicentred studies are needed to enable reasonable decisions to be made about tacrine's clinical significance and potential utility. In the interim there will be "inconclusive, contradictory trials of small size from which it is almost impossible to make generalisations for clinical practice. "4 LON S SCHNEIDER
Department of Psychiatry, University of Southern California, Los Angeles, California 90033, United States I Chatellier G, Lacomblez L on behalf ot(iroupe liraintais d'I(tude de la Tetrahydroaminoacridinie. racrine (tetrahvdroaminoacridine; rHA) and lecithin in senile dementia of the Alzheimer type: a multicentrc trial. BreIed] 1990;300:495-9.
(24 February.) 2 Cohen J. Statistical ptrwer analvsis Jor the behavioral sciences. Hillsdale, New Jersey: Lawrence Erlbaum Associates, 1988. 3 Gauthier S, Bouchard R, Bacher Y, et al. Progress ots the Canadian multicenter trial of tetrahydroanminoacridinc ssith
lecithin in Alzhcimer's disease. Can 16: 543-6. 4
Goodmats NW. Multicetntre trials.
Br
J7 \eueol
ScI 1989;
tIed] 1990;300:538. (24
Fcbruiary.)
Pocket programmable computers in clinical practice SIR,-Dr Jeremy Wyatt commented on the potential value of putting clinical scoring systems on a pocket sized programmable computer.' We have
BMJ VOLUME 300
7 APRIL 1990
been developing such systems for the management of a varietv of medical conditions, of which a system for the diagnosis, physical examination, and treatment of hvpercholesterolaemia is currentlv undergoing field trials with general practitioners in the west midlands. The system is supplied for the use of a Psion Organiser and incorporates regional knowledge in a particular subject as appropriate for use in primary care or by junior hospital medical staff. The medical knowledge supplied facilitates patient management up to a level considered appropriate for the general practitioner or non-specialist physician and contains locally determined referral criteria for specialist attention. Preliminary results suggest that such a format is acceptable to both general practitioners and practice nurses and that programmable pocket computers are an appropriate vehicle for the devolution of medical expertise to nonspecialists or non-medical staff such as practice nurses. Inappropriate referrals to hospital clinics mav also be reduced by providing clear guidelines to referring general practitioners with this medium. We also envisage using such systems for providing useful advice on foreign travel prophylaxis, poisoning, laboratory sampling requirements, protocols for standard medical tests, etc. The potential for devolution of medical expertise is just beginning to be realised. Pocket sized computers can retain and instantaneously recall a large amount of medically relevant data and, unlike a text book, can be readily updated, are portable, and have a variety of other functions. M F RYAN
N CORBETT
I CLARK
M PETERS
Department of ledical Computing, Wolfson Research Laboratories, Queen Elizabeth Medical Centre, Birtningham B15 2FH 1 Wyatt J. Construction of clinical scoring systems. Br Med J
1990;300:538-9. (24 February-.'l
MIF, Clark I, P'etcrs .Ml, Cramb R. A low cost indicator for rationalisation of lipid clinic refcrrals. In: Halloran SP, ed. Focus '89. Londot: Association of Clinical Biochemists,
2 Ryan
1989:68.
Streptokinase in acute aortic dissection SIR,-The report by Mr J Butler and colleagues emphasises the risk of leaving major medical management decisions to inexperienced junior medical staff.' In each of the cases described the house officer was apparently the first medical practitioner to see patients admitted as emergencies. This is an illogical method for assessing patients. The administration of potentially life saving but also potentially life threatening treatments surely requires a higher level of skill and experience than should be expected from a house officer. It is noteworthy that in all of the patients described in this report aortic dissection should have been considered on the basis of the chest radiograph, which was available to the admitting house officer. There is insufficient data in the report by Dr Nicholas P Curzen and colleagues2 to judge whether the same circumstances occurred in their patients, although in one case the chest radiograph was suggestive of aortic dissection. In the study by Mr Butler and colleagues it is not surprising that the diagnosis of aortic dissection was not considered by the admitting house officer, both in view of the limited experience in cardiovascular diseases that most medical students receive during their training and also the limited exposure to training in radiology in most undergraduate medical training. The problems concerned in distinguishing acute aortic dissection from other causes of severe acute chest pain should not cause the uses of thrombolytic agents to be reduced. Rather it is
BMJ
VOLUME
300
7
APRIL
1990
an important reason for changing the way in which patients admitted with acute conditions are assessed. Surely it is no longer satisfactory to delegate assessment of acutely ill patients to the most junior members of the medical team; this should now be the responsibility of a more senior (that is, registrar grade) admitting physician. This will improve the accuracy of the initial management of these patients and will help to prevent the potential catastrophes described by Mr Butler and colleagues. It may also help to prevent an epidemic of unnecessary requests for investigation for possible aortic dissection before doctors administer thrombolytic treatment. G HARTNELL
Department of Radiodiagnosis, Bristol Royal Infirmary, Bristol BS2 8HW I Butler J, Davies AH, Wecstab- S. Streptokinase in acute aortic dissection. BrMteedJ 1990;300:517-9. (24 February.) 2 Curzen NP, Clarke B, Gray HH. Intravenous thrombolysis for suspected myocardial infarction: a cautionary note. Br MedJ7 1990;300:513. (24 February.)
Medical problems of adults who were sexually abused in childhood SIR,-Assessing the possible effects of adversity during childhood on health in adulthood requires studies that either include appropriate control subjects for comparison or eliminate the effects of confounding variables statistically. In their study on the medical problems of adults who were sexually abused in childhood Dr R P Arnold and colleagues used neither of these techniques.' This leads to most basic problems in interpreting their results. For example, we would surely expect the total number of medical specialties contacted by any person (whether sexually abused in childhood or not) to increase with age. The data presented (tables I and III) show exactly this expected correlation between age and the number of specialties contacted for the seven abused women (Spearman's test, Q=088; z=2 16; p
