- Email: [email protected]
Unlocking the datasets
7
Petersen P, 323: 482.
8
Antiplatelet Trialists’ Collaboration. Secondary prevention of vascular disease by prolonged antiplatelet treatment. BMJ 1988; 296: 320-31. Kopecky SL, Gersh BJ, McGoon MD, et al. The natural history of lone atrial fibrillation. A population-based study over three decades. N Engl J Med 1987; 317: 669-74. Boysen G, and the Atrial Fibrillation Investigators. Risk factors for
9
10
Boysen G.
Stroke in atrial fibrillation. N
Engl J Med 1990;
stroke and effect of warfarin in non-valvular atrial fibrillation: a pooling of data from five clinical trials. Neurology 1993; 43 (suppl
A):
390.
11
12
Bussey HI, Force RW, Bianco TM, Leonard AD. Reliance on prothrombin time ratios causes significant errors in anticoagulation therapy. Arch Intern Med 1992; 152: 278-82. Laupacis A, Albers G, Dunn M, Feinberg W. Antithrombotic therapy in atrial fibrillation. Chest 1992; 102: 426S-33S.
Magnetic
resonance
for detection of
abnormalities in partial epilepsy resonance imaging is helpful in the detection of brain lesions in patients with intractable complex partial seizures. In most adults with this disorder and clinical evidence of a temporal lobe focus, MR shows hippocampal sclerosis (HS),1,2 and the response to surgery is good.1 In children, HS may be less frequent.3,4 The cause/effect relation between HS and seizures has not been firmly established. Cross and colleagues5 lately reported their findings in 30 children aged 0-9-15years with intractable complex partial seizures. Seizures were categorised clinically as temporal, extratemporal, or unlocalised; all patients underwent MR imaging, and 29 of the 30 underwent electroencephalography with surface electrodes. The researchers’ conclusions were: (a) MR is highly sensitive for the presence of structural abnormality (much more so than conventional electroencephalography); (b) intractable temporal lobe epilepsy is almost always associated with an abnormality visible on MR, most commonly HS even in this paediatric series; (c) serial MR studies may in future clarify the cause/effect relation between HS and temporal lobe epilepsy; (d) HS is strongly associated with early status epilepticus; (e) widespread cortical abnormality is common in this group; and (/) MR may lead to earlier surgical intervention. In the series of Cross et al, HS was present in 62 °o of patients with temporal or unlocalised seizures. This figure is similar to the frequency in adult seriesbut differs significantly from the results of Hirschorn et aJ,4 who saw no hippocampal atrophy on MR in any of 20 children with temporal lobe epilepsy (though gliosis was found in most
Magnetic
surgical specimens). In patients with complex partial seizures, the MR sequences used must optimise delineation of hippocampal volume, morphology, and signal intensity. Cross et al used a heavily Tl-weighted IR sequence and a dual-echo T2weighted STIR sequence, imaging in axial and coronal planes orthogonal to the hippocampus. A small field of view and very thin contiguous images with three-dimensional technique would be even more likely to detect abnormalities2.7 than the 5 - 0 mm thick images with 2.5mm inters lice gap used by Cross et al. Phased-array surface coils may achieve ideal resolution’ but are not routinely available.
Quantitative assessment of hippocampal volumes is recommended by several researchers1,2,4 and the technical aspects have been well described. I therefore wonder why 1252
objective measurements were not used here. Nevertheless, the blinded subjective analysis of MR images carried out by Cross and colleagues shows good correlation with clinical findings: all 19 children with clinical temporal lobe epilepsy had unilateral (14) or bilateral (5) temporal abnormalities. None of the 4 patients with a clinical diagnosis of extratemporal epilepsy showed hippocampal or temporal abnormality on MR. However, clinical left-right lateralisation was possible in only 11 of the 30 children. The researchers do not state whether conflicting lateralisation of clinical and MR findings occurred in any of these 11; false lateralisation by MR is an important error.’ A major weakness of this study is that pathological proof was obtained in only 2 patients-1 with HS and the other with temporal astrocytoma. This lack of pathological corroboration is especially noteworthy in the large group of patients reported as having "widespread cortical changes" (anomaly or injury), seen in 47% of the MR studies (the example shown in one of the figures provides little evidence to substantiate this claim). Since 11 of the 14 children with widespread abnormality on MR also showed a focal lesion which could have accounted for their seizures, it is difficult to correlate all the MR and clinical findings in this group. As Cross et al note, these patients might be expected to respond more poorly to temporal lobe surgery. In a previous paediatric surgical series3 a developmental anomaly was found in the temporal lobe in 10 of 16 patients, so a high prevalence of congenital anomalies is not unexpected in this latest series. However, the prevalence of "widespread abnormalities" is unusual. Carla J Wallace Department of Radiological Sciences, Foothills Hospital, Calgary, Alberta, Canada
Jack CR, Sharbrough FW, Twomey CK, et al. Temporal lobe seizures: lateralization with MR volume measurements of the hippocampal formation. Radiology 1990; 175: 423-29. 2 Cook MJ, Fish DR, Shorvon SD, Straughan K, Stevens JM. Hippocampal volumetric and morphometric studies in frontal and temporal lobe epilepsy. Brain 1992; 115: 1001-15. 3 Duchowny M, Levin B, Jayakar P, et al. Temporal lobectomy in early childhood. Epilepsia 1992; 33: 298-303. 4 Hirschorn KA, Jack CR, Parisi JE, Marsh WR, Sharbrough FW, Cascino GD. Correlation of MRI hippocampal volume measurements and pathology in intractable temporal lobe epilepsy in children. Neurology 1991; 41 (S): 170. 5 Cross JH, Jackson GD, Neville BGR, et al. Early detection of abnormalities in partial epilepsy using magnetic resonance. Arch Dis Child 1993; 69: 104-09. 6 Jackson GD, Berkovic SF, Tress BM, Kalnins RM, Fabinyi GCA, Bladin PF. Hippocampal sclerosis can be reliably detected by magnetic resonance imaging. Neurology 1990; 40: 1869-75. 7 Tsuruda JS, Hayes CE, Alvord EC, Ojemann LM, Ojemann GA. A phased array RF coil assembly for high resolution fast spin echo imaging of mesial temporal sclerosis: correlation with histology. ASNR Meeting Notes, Vancouver, 1992: 174-75. 1
Unlocking the datasets How can a physician make a rational decision to start an individual patient on a new therapy that is both expensive and of uncertain efficacy? As a result of swift dissemintaion of encouraging preliminary findings via word of mouth or a patients’ organisation, doctors may be requested to initiate treatments that would severely dent the payers’ budget. For example, in Gaucher’s disease treatment with a drug costing 100000 ($150000 dollars) a year has been proposed,! but published clinical data on the effects of therapy are sparse, and there has been no randomised clinical trial.
all the convictions had been wrongful, while on some occasions the accused soldier had virtually no defence to the charges laid against him. Many other cases, however, occasion considerable disquiet. The courts martial took place in accordance with the ordinary principles of the British criminal law. No soldier, therefore, should have been convicted unless the court was satisfied that, at the time the offence was committed, he had had the necessary mens rea or knowledge of wrongdoing. If the evidence showed that his mind might have been confused because of the shocks or stresses of the battlefield it must be doubtful if he was mentally in control of his actions. The British Army in 1914 was almost totally ignorant of battle neurosis. This was inexcusable. It was well known in medical circles that during the Russo-Japanese war of 1904-05 the Russians had set up a special forward clearing hospital for psychiatric casualties. An American medical officer, R L Richards, who had studied the results of this experiment reported in 1910 that in a future war the opposing armies would sustain "a larger percentage of mental diseases" than had ever been known before.2 Throughout the 1914-18 war the idea persisted among the senior ranks of the British Army-and among military medical officers-that the condition then known as "shellshock" was closely akin to a state of funk. Lord Moran, from his experiences as a medical officer in that war, knew that shell-shock was a genuine illness but his views were not generally shared by his colleagues in France or back home.3 The distinguished neurologist W H R Rivers, when he was treating Siegfried Sassoon (himself an officer of courage and conscience) at the Craiglockhart War Hospital in 1917, told him that the most senior local army doctor had asserted that "he never had and never would recognise the existence Mark McCarthy of such a thing as shell-shock".4 In France British military Department of Public Health, Camden and Islington Health Authority, London, UK doctors were not merely content to discount mental stress as a factor; on occasion they seemed positively determined on a 1 Cox TM. Gaucher’s disease: a brand leader. Lancet 1993; 342: 694-95. death sentence.s 2 Chalmers I. Randomised clinical trials. N Engl J Med 1991; 325: 1514. The War Office Committee of Inquiry into shell-shock 3 Advisory Group on Health Technology Assessment. Assessing the reported in 1922 that the term "shell-shock" had been effects of health technologies: principles, practice, proposals. London: of 1993. Health, "a gross and costly misnomer". The war had produced no Department 4 Hellman S, Hellman DS. Of mice but not of men: problems of the new nervous disorders, they said. The psychiatric randomised clinical trial. N Engl J Med 1991; 324: 1585-89. casualties had been of three types, the report said: genuine 5 Editorial. Databases for health care outcomes. Lancet 1989; ii: 195-96. concussion caused by an explosion, emotional shocks, and 6 Jennett B. High technology medicine: benefits and burdens. London: Nuffield Provincial Hospitals Trust, 1984. nervous exhaustion. Of the first category, Lord Moran said that he had known brave men whose brains were damaged by the blast of a shell so that they were left "with their skins intact but with dishevelled minds".3 Some soldiers executed for desertion had never been in A stain on army medicine action: it is those who had deserted or committed acts of Last month in the House of Commons, Mr Andrew cowardice in the front line whose cases should now be re-examined. I could give many examples of convictions Mackinlay introduced a Private Member’s Bill seeking pardons for British soldiers executed for purely military which, as a retired judge and former soldier, I find offences during the 1914-18 war. He proposed either a disturbing, but one will illustrate my point. A lanceblanket pardon for all the men or that every case should be sergeant went absent from his battalion in 1914 just after the reviewed individually by a tribunal of High Court judges. retreat from Mons. A French civilian described him as read executed all of the files on the men 1918 Having (the looking exhausted, and noticed that he had a flesh wound on ones will not be public until January) when compiling the one hand. Three months later he was arrested and earlier edition of my book,1 I wholeheartedly support the interviewed by a major in the Military Police who reported second proposal, except that the tribunal should include at that he appeared to be half-dazed and was unable or least one psychiatrist and a senior army officer, besides unwilling to answer the simplest questions. At his courtmartial the soldier said that he had undergone some sort of judges. Between 1914 and 1918 a total of 266 men were shot for nervous breakdown directly after being wounded-but he and 27 for cowardice or cowardice-associated was given no sort of neurological examination before being desertion, offences. Although I deplore the fact that any of these men shot. These soldiers were not convicted by the ordinary were sentenced to death, a blanket pardon might imply that criminal courts. They appeared on capital charges before
Some would say that all
patients starting
unproven
randomised clinical trial.2 In the UK, the Department of Health’s strategy for technology assessment3 suggests that randomised controlled trials are needed "when technologies have only moderate (but still worthwhile) effects" and the effects of confounding become more important. Yet recruitment into a trial may not be achievable for various reasons: some patients refuse randomisation; some do not fall within the criteria set for the trial, some are withdrawn for clinical or social reasons; and for some physicians, the ethical problems of randomised assignment and informed consent are insurmountable.4 While historical controls are unsatisfactory because of patient selection, large datasets can yield worthwhile information.5 Jennett has shown that systematic collection of clinical datasets can indicate preferred methods of management and intervention.6 The dataset should record evidence of the drug’s effect on all patients under treatment. Subanalyses of groups stratified by severity, age, or other relevant factors can help physicians "match" their patient and predict the potential outcome. At present, manufacturers guard the results of their in-house studies: they must be persuaded to be more open. Patients and physicians should insist that data derived from their care be made available (with full confidentiality protected) for other physicians to draw on. Insurance agencies and purchasers should make professional access to the clinical datasets of innovative treatment a necessary condition before payment of the treatment is agreed. A new openness-unlocking the datasets-would enhance costeffective patient care.
treatment ought be entered into
a
1253
Petersen P, 323: 482.
8
Antiplatelet Trialists’ Collaboration. Secondary prevention of vascular disease by prolonged antiplatelet treatment. BMJ 1988; 296: 320-31. Kopecky SL, Gersh BJ, McGoon MD, et al. The natural history of lone atrial fibrillation. A population-based study over three decades. N Engl J Med 1987; 317: 669-74. Boysen G, and the Atrial Fibrillation Investigators. Risk factors for
9
10
Boysen G.
Stroke in atrial fibrillation. N
Engl J Med 1990;
stroke and effect of warfarin in non-valvular atrial fibrillation: a pooling of data from five clinical trials. Neurology 1993; 43 (suppl
A):
390.
11
12
Bussey HI, Force RW, Bianco TM, Leonard AD. Reliance on prothrombin time ratios causes significant errors in anticoagulation therapy. Arch Intern Med 1992; 152: 278-82. Laupacis A, Albers G, Dunn M, Feinberg W. Antithrombotic therapy in atrial fibrillation. Chest 1992; 102: 426S-33S.
Magnetic
resonance
for detection of
abnormalities in partial epilepsy resonance imaging is helpful in the detection of brain lesions in patients with intractable complex partial seizures. In most adults with this disorder and clinical evidence of a temporal lobe focus, MR shows hippocampal sclerosis (HS),1,2 and the response to surgery is good.1 In children, HS may be less frequent.3,4 The cause/effect relation between HS and seizures has not been firmly established. Cross and colleagues5 lately reported their findings in 30 children aged 0-9-15years with intractable complex partial seizures. Seizures were categorised clinically as temporal, extratemporal, or unlocalised; all patients underwent MR imaging, and 29 of the 30 underwent electroencephalography with surface electrodes. The researchers’ conclusions were: (a) MR is highly sensitive for the presence of structural abnormality (much more so than conventional electroencephalography); (b) intractable temporal lobe epilepsy is almost always associated with an abnormality visible on MR, most commonly HS even in this paediatric series; (c) serial MR studies may in future clarify the cause/effect relation between HS and temporal lobe epilepsy; (d) HS is strongly associated with early status epilepticus; (e) widespread cortical abnormality is common in this group; and (/) MR may lead to earlier surgical intervention. In the series of Cross et al, HS was present in 62 °o of patients with temporal or unlocalised seizures. This figure is similar to the frequency in adult seriesbut differs significantly from the results of Hirschorn et aJ,4 who saw no hippocampal atrophy on MR in any of 20 children with temporal lobe epilepsy (though gliosis was found in most
Magnetic
surgical specimens). In patients with complex partial seizures, the MR sequences used must optimise delineation of hippocampal volume, morphology, and signal intensity. Cross et al used a heavily Tl-weighted IR sequence and a dual-echo T2weighted STIR sequence, imaging in axial and coronal planes orthogonal to the hippocampus. A small field of view and very thin contiguous images with three-dimensional technique would be even more likely to detect abnormalities2.7 than the 5 - 0 mm thick images with 2.5mm inters lice gap used by Cross et al. Phased-array surface coils may achieve ideal resolution’ but are not routinely available.
Quantitative assessment of hippocampal volumes is recommended by several researchers1,2,4 and the technical aspects have been well described. I therefore wonder why 1252
objective measurements were not used here. Nevertheless, the blinded subjective analysis of MR images carried out by Cross and colleagues shows good correlation with clinical findings: all 19 children with clinical temporal lobe epilepsy had unilateral (14) or bilateral (5) temporal abnormalities. None of the 4 patients with a clinical diagnosis of extratemporal epilepsy showed hippocampal or temporal abnormality on MR. However, clinical left-right lateralisation was possible in only 11 of the 30 children. The researchers do not state whether conflicting lateralisation of clinical and MR findings occurred in any of these 11; false lateralisation by MR is an important error.’ A major weakness of this study is that pathological proof was obtained in only 2 patients-1 with HS and the other with temporal astrocytoma. This lack of pathological corroboration is especially noteworthy in the large group of patients reported as having "widespread cortical changes" (anomaly or injury), seen in 47% of the MR studies (the example shown in one of the figures provides little evidence to substantiate this claim). Since 11 of the 14 children with widespread abnormality on MR also showed a focal lesion which could have accounted for their seizures, it is difficult to correlate all the MR and clinical findings in this group. As Cross et al note, these patients might be expected to respond more poorly to temporal lobe surgery. In a previous paediatric surgical series3 a developmental anomaly was found in the temporal lobe in 10 of 16 patients, so a high prevalence of congenital anomalies is not unexpected in this latest series. However, the prevalence of "widespread abnormalities" is unusual. Carla J Wallace Department of Radiological Sciences, Foothills Hospital, Calgary, Alberta, Canada
Jack CR, Sharbrough FW, Twomey CK, et al. Temporal lobe seizures: lateralization with MR volume measurements of the hippocampal formation. Radiology 1990; 175: 423-29. 2 Cook MJ, Fish DR, Shorvon SD, Straughan K, Stevens JM. Hippocampal volumetric and morphometric studies in frontal and temporal lobe epilepsy. Brain 1992; 115: 1001-15. 3 Duchowny M, Levin B, Jayakar P, et al. Temporal lobectomy in early childhood. Epilepsia 1992; 33: 298-303. 4 Hirschorn KA, Jack CR, Parisi JE, Marsh WR, Sharbrough FW, Cascino GD. Correlation of MRI hippocampal volume measurements and pathology in intractable temporal lobe epilepsy in children. Neurology 1991; 41 (S): 170. 5 Cross JH, Jackson GD, Neville BGR, et al. Early detection of abnormalities in partial epilepsy using magnetic resonance. Arch Dis Child 1993; 69: 104-09. 6 Jackson GD, Berkovic SF, Tress BM, Kalnins RM, Fabinyi GCA, Bladin PF. Hippocampal sclerosis can be reliably detected by magnetic resonance imaging. Neurology 1990; 40: 1869-75. 7 Tsuruda JS, Hayes CE, Alvord EC, Ojemann LM, Ojemann GA. A phased array RF coil assembly for high resolution fast spin echo imaging of mesial temporal sclerosis: correlation with histology. ASNR Meeting Notes, Vancouver, 1992: 174-75. 1
Unlocking the datasets How can a physician make a rational decision to start an individual patient on a new therapy that is both expensive and of uncertain efficacy? As a result of swift dissemintaion of encouraging preliminary findings via word of mouth or a patients’ organisation, doctors may be requested to initiate treatments that would severely dent the payers’ budget. For example, in Gaucher’s disease treatment with a drug costing 100000 ($150000 dollars) a year has been proposed,! but published clinical data on the effects of therapy are sparse, and there has been no randomised clinical trial.
all the convictions had been wrongful, while on some occasions the accused soldier had virtually no defence to the charges laid against him. Many other cases, however, occasion considerable disquiet. The courts martial took place in accordance with the ordinary principles of the British criminal law. No soldier, therefore, should have been convicted unless the court was satisfied that, at the time the offence was committed, he had had the necessary mens rea or knowledge of wrongdoing. If the evidence showed that his mind might have been confused because of the shocks or stresses of the battlefield it must be doubtful if he was mentally in control of his actions. The British Army in 1914 was almost totally ignorant of battle neurosis. This was inexcusable. It was well known in medical circles that during the Russo-Japanese war of 1904-05 the Russians had set up a special forward clearing hospital for psychiatric casualties. An American medical officer, R L Richards, who had studied the results of this experiment reported in 1910 that in a future war the opposing armies would sustain "a larger percentage of mental diseases" than had ever been known before.2 Throughout the 1914-18 war the idea persisted among the senior ranks of the British Army-and among military medical officers-that the condition then known as "shellshock" was closely akin to a state of funk. Lord Moran, from his experiences as a medical officer in that war, knew that shell-shock was a genuine illness but his views were not generally shared by his colleagues in France or back home.3 The distinguished neurologist W H R Rivers, when he was treating Siegfried Sassoon (himself an officer of courage and conscience) at the Craiglockhart War Hospital in 1917, told him that the most senior local army doctor had asserted that "he never had and never would recognise the existence Mark McCarthy of such a thing as shell-shock".4 In France British military Department of Public Health, Camden and Islington Health Authority, London, UK doctors were not merely content to discount mental stress as a factor; on occasion they seemed positively determined on a 1 Cox TM. Gaucher’s disease: a brand leader. Lancet 1993; 342: 694-95. death sentence.s 2 Chalmers I. Randomised clinical trials. N Engl J Med 1991; 325: 1514. The War Office Committee of Inquiry into shell-shock 3 Advisory Group on Health Technology Assessment. Assessing the reported in 1922 that the term "shell-shock" had been effects of health technologies: principles, practice, proposals. London: of 1993. Health, "a gross and costly misnomer". The war had produced no Department 4 Hellman S, Hellman DS. Of mice but not of men: problems of the new nervous disorders, they said. The psychiatric randomised clinical trial. N Engl J Med 1991; 324: 1585-89. casualties had been of three types, the report said: genuine 5 Editorial. Databases for health care outcomes. Lancet 1989; ii: 195-96. concussion caused by an explosion, emotional shocks, and 6 Jennett B. High technology medicine: benefits and burdens. London: Nuffield Provincial Hospitals Trust, 1984. nervous exhaustion. Of the first category, Lord Moran said that he had known brave men whose brains were damaged by the blast of a shell so that they were left "with their skins intact but with dishevelled minds".3 Some soldiers executed for desertion had never been in A stain on army medicine action: it is those who had deserted or committed acts of Last month in the House of Commons, Mr Andrew cowardice in the front line whose cases should now be re-examined. I could give many examples of convictions Mackinlay introduced a Private Member’s Bill seeking pardons for British soldiers executed for purely military which, as a retired judge and former soldier, I find offences during the 1914-18 war. He proposed either a disturbing, but one will illustrate my point. A lanceblanket pardon for all the men or that every case should be sergeant went absent from his battalion in 1914 just after the reviewed individually by a tribunal of High Court judges. retreat from Mons. A French civilian described him as read executed all of the files on the men 1918 Having (the looking exhausted, and noticed that he had a flesh wound on ones will not be public until January) when compiling the one hand. Three months later he was arrested and earlier edition of my book,1 I wholeheartedly support the interviewed by a major in the Military Police who reported second proposal, except that the tribunal should include at that he appeared to be half-dazed and was unable or least one psychiatrist and a senior army officer, besides unwilling to answer the simplest questions. At his courtmartial the soldier said that he had undergone some sort of judges. Between 1914 and 1918 a total of 266 men were shot for nervous breakdown directly after being wounded-but he and 27 for cowardice or cowardice-associated was given no sort of neurological examination before being desertion, offences. Although I deplore the fact that any of these men shot. These soldiers were not convicted by the ordinary were sentenced to death, a blanket pardon might imply that criminal courts. They appeared on capital charges before
Some would say that all
patients starting
unproven
randomised clinical trial.2 In the UK, the Department of Health’s strategy for technology assessment3 suggests that randomised controlled trials are needed "when technologies have only moderate (but still worthwhile) effects" and the effects of confounding become more important. Yet recruitment into a trial may not be achievable for various reasons: some patients refuse randomisation; some do not fall within the criteria set for the trial, some are withdrawn for clinical or social reasons; and for some physicians, the ethical problems of randomised assignment and informed consent are insurmountable.4 While historical controls are unsatisfactory because of patient selection, large datasets can yield worthwhile information.5 Jennett has shown that systematic collection of clinical datasets can indicate preferred methods of management and intervention.6 The dataset should record evidence of the drug’s effect on all patients under treatment. Subanalyses of groups stratified by severity, age, or other relevant factors can help physicians "match" their patient and predict the potential outcome. At present, manufacturers guard the results of their in-house studies: they must be persuaded to be more open. Patients and physicians should insist that data derived from their care be made available (with full confidentiality protected) for other physicians to draw on. Insurance agencies and purchasers should make professional access to the clinical datasets of innovative treatment a necessary condition before payment of the treatment is agreed. A new openness-unlocking the datasets-would enhance costeffective patient care.
treatment ought be entered into
a
1253