Mortality from epilepsy: results from a prospective population-based study

Ruggiero HA, Castellanos H, Caprissi LF, Caprissi LT. Heparin effect blood viscosity. Clin Cardiol 1982; 5: 215-18. 14 Second European consensus document on chronic critical leg ischaemia. Circulation 1991; 84 (suppl 4): 1-26. 15 Underwood MJ, More RS. The aspirin papers. BMJ 1994; 308: 71-72. 16 Antiplatelet Trialists Collaboration. Collaborative overview of randomised trials of antiplatelet therapy-II; maintenance of vascular graft or arterial patency by antiplatelet therapy. BMJ 1994; 308: 13

on

159-68. 17 Au YPT,

Montgomery KF,

Clowes AW.

Heparin inhibits collagenase

ester responsive element in primate arterial smooth muscle cells. Circ Res 1992; 70: 1062-69. 18 Rosenberg RD. Vascular smooth muscle cell proliferation: basic investigations and new therapeutic approaches. Thromb Haemost 1993; 70: 10-16.

expression mediated by phorbol

19

20

Mackey WC, McCullough JL, Conlon TP, et al. The costs of surgery for limb-threatening ischaemia. Surgery 1986; 99: 26-35. Forbes JF. Economics of prevention and treatment. In: Fowkes FGR, ed. Epidemiology of peripheral vascular disease. Berlin: Springer-Verlag, 1991. Pp 235-43.

Mortality from epilepsy: results from based study

increased risk Patients with epilepsy may be subject to of premature death from the underlying cause, or from the epilepsy intself. The extent and nature of this risk has been insufficiently investigated. Standard mortality ratios (SMRs) of patients with newly diagnosed epilepsy were determined in a prospective national population-based study. 1091 patients with newly diagnosed or suspected epilepsy were ascertained who were attending one of 275 UK general practices from 1984-1987. 1091 patients were classified after 6 months as definite epilepsy (564), possible epilepsy (228), febrile seizures (220), or not epilepsy (79). Over a median follow up of 6·9 years the SMR for patients with definite or possible epilepsy was 2·5 (95% Cl 2·1-2·9), and 3·0 (2·5-3·7) for definite epilepsy. The SMR was highest during the first year after diagnosis 5·1 (3·8-6·5), declined to 2·5 (1·5-3·9) at 3 years, and 1·3 (0·7-2·0) at 5 years. The commonest causes of death were pneumonia (SMR 7·2), cancer (3·5), and stroke (3·7). The SMR for patients with idiopathic epilepsy was 1·6 (1·0-2·4), remote symptomatic epilepsy 4·3 (3·3-5·5), and acute symptomatic epilepsy 2·9 an

(1·7-4·5). Mortality in patients with newly-diagnosed epilepsy was high, mainly due to the underlying cause. The SMR for idiopathic epilepsy was also raised, suggesting that epilepsy per se may carry a small risk of death.

Epilepsy Research Group, National Hospital for Neurology and Neurosurgery and Institute of Neurology, London, and Chalfont Centre for Epilepsy, Chalfont St Peter, Bucks SL9 ORJ, UK (O C Cockerell MRCP, Y M Hart MRCP, J W A S Sander MD, D M G Goodridge FRCGP, S D Shorvon FRCP); and MRC Biostatistics Unit, Institute of Public Health, Cambridge (A L Johnson PhD)

918

prospective population-

Introduction

Summary

Correspondence to:

a

Dr O C Cockerell

Gowers observed in 1885 that "the danger to life of patients with epilepsy is not great",’ an opinion widely held since,2,3 although it is now accepted that patients with epilepsy have a higher death rate than the general population.4-9 The extent of the increased risk of death (especially to patients with idiopathic epilepsy) and its causes are unknown. The National General Practice Study of Epilepsy (NGPSE) is a population-based prospective study that identified patients with a diagnosis of possible epilepsy at presentation, and that has now followed their course for up to 9 years after diagnosis. The study has been possible because the general practice system in the UK, whereby every patient has a unique National Health Service identification number and can be traced through the Central Register, allows notification of deaths occurring in the NGPSE study even if the patient has moved away from the general practitioner (GP) who originally registered the patient.

Patients and methods methodology of NGPSE has been reported previously.’"’" In summary, patients with newly diagnosed epilepsy (or possible epilepsy) were identified from 275 general practices throughout the UK. Information received from GPs and hospital doctors was submitted to a diagnostic review panel 6 months after registration. Epilepsy was defined by the occurrence of one or more seizures, whatever the cause. Neonatal seizures or patients previously diagnosed as having epilepsy were excluded. 1091 patients were included, 564 patients classified as definite epilepsy, 228 as possible epilepsy, 220 with febrile convulsions, and 79 who did not have epilepsy. Data were obtained on seizure status, medical care, and treatment by means of an annual follow-up form to the GP. Patients were followed prospectively from the index seizure (the seizure which led to identification of epilepsy), and were flagged by the National Health Service Central Register, so we were informed if any patient died. Cause of death was obtained from the death certificate, and supplementary information from the GP and hospital records. Deaths were assigned an International Classification of Diseases (ICD)-9 cause. Analysis was by the person-years method,’2 and summarised by standardised mortality ratios (SMRs) with 95% CIs, and twotailed significance tests. Expected numbers of deaths were calculated for all causes, and for specific causes of death, by sex The

*Significance test based on Poisson distribution for SRM Table 1: All-cause mortality

different from 1

(2-sided).

and age groups (0, 1, 2, 3, 4, 5-9, 10-14, 85-89, 90-100 years) and year-specific (1984 to 1992) death rates for the population of England and Wales. SMRs were calculated for the whole cohort and also for patients with definite epilepsy, possible epilepsy, and febrile seizures. Analysis of mortality was stratified by the interval from the first to the index seizure. SMRs were also calculated for patients with definite epilepsy who were classified as remote symptomatic (eg, brain tumours and vascular disease), acute symptomatic (eg, alcoholic and metabolic), idiopathic, and causes due to a neurological deficit acquired at birth. Patients ...,

identified between 1984 and 1987 and followed until death, or if still alive, to December 31, 1992.

were

Results (25th,

Median

75th

centiles) follow up

was

6-9

(5-8-7-8)

years, equivalent to over 7500 person-years. The number of person-years was similar for the different classifications of epilepsy. 161 people (75 males, 86 females) died, compared with 69 deaths from all-causes that would have been expected (table 1). The all cause SMR in the whole cohort was 2-3 (95% CI 1-9-2-7, p<0-001), for patients with definite epilepsy 3-0 (2-5-3-7), and possible epilepsy 1-6 (1-2-2-2). There were no deaths in the group with

febrile seizures. Mortality within selected age groups is summarised in table 2. Deaths in 16 people aged under 50 were caused by brain tumours in 5, injuries (ICD codes 800-999) in 5, and myocardial infarction, subarachnoid haemorrhage, cerebral degeneration, viral pneumonia, and Huntington’s disease in 1 each. Mortality for patients with definite epilepsy, and definite and possible epilepsy, for each year of follow up is shown in table 3. The SMR was highest at 6-6 in patients with definite, and 5-1 in possible and definite epilepsy, during the first year of follow up, approximately halved during the subsequent 3 years, and then halved again after 4 years to an SMR of around 1 after 4-5 years. Similar results were obtained with the period of exposure to seizures (defined as the interval from first seizure); among

those with a history of seizures for less than 2 years, the SMR was 5-1 (4-0-6-3), and then declined to 2-3 (1-5-3-2) in those having a history of at least 2 and up to 4 years, and to 1-3 (0-9-1-7) in those with a history of more than 4 years. For those with possible or definite epilepsy, when the index seizure was the first seizure, the overall SMR was 3-3 (2-6-4-1), 2-7 (1-8-3-9) when the first seizure occurred within 6 months of the index seizure, and 1-4 (0-9-2-1) when it occurred 6 months after the index seizure. The all-cause mortality for patients with definite epilepsy classified by aetiology is shown in table 4. Mortality from selected causes of death in those cases with definite epilepsy, and with possible or definite epilepsy, is shown in table 5. For those with definite seizures only, the cause-specific SMRs were higher and the CIs wider. Mortality from all causes other than those selected in table 5 was slightly above that in the standard population (SMR 1-4 [1’0-1’9]; p<0-025) but the higher risk was confined to women (men p>0-95, women p<0-001), even in those with definite epilepsy (men 0-95 [0-48-1-7], women 2-3 [1-4-35]). Only 2 patients had epilepsy-related deaths, 1 drowned in the bath and the other suffered severe burns

Table 3: Mortality for each year of follow up from index seizure in patients with definite epilepsy and definite or possible

epilepsy

Table 2: All cause mortality by age-group in patients with definite epilepsy and definite or possible epilepsy

Table 4: All-cause mortality in according to aetiology

patients with definite epilepsy

919

first seizure, and so the death rate may be underestimated because patients could have been excluded who had their first seizure in the same time period, but who had died before ascertainment. We examined this by calculating SMRs for patients according to whether the interval between first seizure and index seizure was within 6 months. The SMR was highest at 3-30 when the index seizure was also the first seizure, 2-73 if the first was within 6 months of the second, and 1-44 if it was not. Part of the explanation is that patients with milder epilepsy such as absence seizures, will consult their doctor later than a patient with seizures secondary to a stroke or brain tumour. Our results may, however, still underestimate the death rate; although if this is so, the effect will be probably more for symptomatic rather than

idiopathic epilepsy. Table 5: Selected causes of death for possible or definite epilepsy

patients with definite

or,

consequent upon a seizure. No patients had sudden unexpected death. 1 patient committed suicide, and there were 3 violent deaths in patients with definite epilepsy.

Discussion Previous studies of mortality of epilepsy were hindered by methodological problems. Some studies did not specify the case definition for epilepsy,8,13 and most were in selected populations, such as institutions/,14 or populations of insured patients.3 The application of such results to a wider, and usually less disabled population is of doubtful validity. Ascertainment has also often been inadequate, particularly so in studies that identified cases from death certificates,2 a method that is unreliable because death certificates often do not state that the patient had epilepsy.6 Another difficulty is the determination of the cause of death when the contribution of epilepsy is subjective. 15 Cross sectional studies, and studies which fail to ascertain new cases of epilepsy, will fail to determine the large initial mortality of epilepsy, and will thus also misrepresent particular causes of death, such as brain tumours. Errors due to biased patient selection can be eliminated by a population-based study. Two previous studies have been carried out in unselected groups of patients;6,s both were retrospective, with the potential to miss patient groups such as the elderly, the very young, and individuals where seizures were not witnessed. Also by only including patients with definite epilepsy, milder cases or those with a delayed diagnosis may be missed. The study design of the NGPSE, by including patients who had definite or possible seizures, allows us to be confident that no patients with epilepsy were missed, even where the diagnosis was not at first clear (an important issue in epilepsy because death rates are highest in the periods immediately after presentation). The NGPSE study includes patients with single seizures who are often not included in the definition of epilepsy, and has already shown that most patients with single seizures go on to experience further seizures," and that there is no pathophysiological difference between patients who have single seizures and those who have more than one. A particular concern in the NGPSE has been the effect of the interval from first seizure to the time of the first medical consultation and registration. We included patients with new-onset epilepsy, not necessarily after the 920

The death rate was highest in the first year (SMR 6-6, [4-8-8-7]), then decreased progressively. Even after 6 years, the SMR was still raised, though not significantly so, at 1-3, findings similar to the only other study to examine mortality rate over time.’ The trend was largely the result of the early death of patients from underlying diseases such as strokes and tumours: the SMR was highest in the remote and acute symptomatic cases because of the underlying lesion, with the majority of these deaths occurring in the first 2 years. The SMR of patients with idiopathic epilepsy was also raised at 1-6, which shows that epilepsy itself carries an increased risk of mortality not related to the underlying cause. The SMR did not fall during follow up. 3 deaths in the idiopathic group were due to an underlying symptomatic aetiology (primary brain tumours which were unrecognised in life), but even with these deaths excluded, the SMR was 1-4. In agreement with other studies we found a raised SMR for cancer, stroke, and pneumonia.5,8,9,14 The SMRs for cancer and strokes were highest in the remote and acute symptomatic groups. The cause of this high cancer risk is unknown. Previous speculations that patients with epilepsy are at risk from anticonvulsant-drug-induced liver cancer lymphomal4,16 were not substantiated as there was only 1 death from liver cancer, and none from lymphoma. It is possible that increased smoking or alcohol intake was to blame, although other studies have not suggested any difference between the smoking habits of people with epilepsy and the general population.9 There has also been speculation that anti-epileptic drugs might influence mortality rate because of their antiarrhythmic effect,5 or by raising plasma lipids." In our study the SMR for ischaemic heart disease was normal for the whole cohort, and only raised in the remote symptomatic group (1-9 [0-8-3-8]), probably due to the association with the high rate of cerebrovascular disease. Pneumonia has been noted as a common cause of death in patients with epilepsy since 1910.’$One hypothesis concerning a possible damaging effect of antiepileptic drugs on pulmonary function has not been substantiated. 19 A high rate of pneumonia deaths was found in all groups in our sample. Evidence from the GP and/or hospital failed to find an underlying cause. The mean age of patients who died of pneumonia was 81-3 years, thus elderly patients with epilepsy may carry an increased susceptibility to pneumonia probably because of higher rates of associated diseases. Suicide has been reported to be more common in patients with epilepsy,20 but this was not confirmed by our

study. Previous studies of suicide in epilepsy have been in patients with chronic intractable epilepsy, whereas the NGPSE incident cases had few severely affected patients. The rate of sudden unexpected death in patients with epilepsy is reported to be between 1:370 and 1: 1100 per year,9 and in patients in the general population is at the upper end of this limit being most prevalent up to the age of 45.Information on sudden unexpected death in epilepsy comes from patients with active epilepsy. In the NGPSE we have so far observed approximately 1000 person years of active epilepsy in persons up to 45 years old so we would not have expected any such deaths in our study, and further follow up may be more informative.

Epilepsy and mortality rate and cause of death. Epilepsia 1974; 15: 191-201. 9 Klenerman P, Sander J, Shorvon S. Mortality in patients with epilepsy: a study of patients in long term residential care. J Neurol Neurosurg Psychiatry 1993; 56: 149-52. 10 Hart YM, Sander JW, Johnson AL, Shorvon SD. National General Practice Study of Epilepsy: recurrence after a first seizure. Lancet 1990; 8

Zielinski JJ.

336: 1271-74.

YM, Sander JW, Shorvon SD. National General Practice Study of Epilepsy and Epileptic Seizures: objectives and study methodology of the largest reported prospective cohort study of epilepsy. National General Practice Study of Epilepsy and Epileptic Seizures (NGPSE). Neuroepidemiology 1989; 8: 221-27. Coleman MP, Herman C, Douglas A. Person-years (PYRS): a further program for cohort study analysis. 1989 (September) WHO. Rodin EA. Mortality. In: The prognosis of patients with epilepsy.

11 Hart

12

13

Illinois: Charles C Thomas. 1968: 156-74.

The NGPSE has received the generous support of Action Research.

14 White

cohort

References 1 Gowers WR. Epilepsy and other chronic disorders: their causes, symptoms, and treatment. London: William Wood & Co, 1885. 2 Schwade E, Owen O. Mortality in epilepsy. JAMA 1954; 156: 1526-27. 3 Livingston S. Living with epileptic seizures. Springfield: Charles C Thomas, 1963. 4 Annegers JF, Elveback LR, Labarthe DR, Hauser WA. Ischaemic heart disease in patients with epilepsy. Epilepsia 1976; 17: 11-14. 5 Annegers JF, Hauser WA, Shirts SB. Heart disease mortality and morbidity in patients with epilepsy. Epilepsia 1984; 25: 699-704. 6 Hauser WA, Annegers JF, Elveback LR. Mortality in patients with epilepsy. Epilepsia 1980; 21: 399-412. 7 Massey EW, Schoenberg BS. Mortality from epilepsy: international patterns and changes over time. Neuroepidemiology 1985; 4: 65-70.

Heat-shock protein 65 and activated T cells in injured arteries

&ggr;/&dgr;

mechanisms are implicated in atherogenesis, we investigated the T-lymphocyte subset and factors related to its activation after acute arterial ligation (22 ligated and 13 non-ligated specimens). Ligated arteries produced heat-shock protein 65 (hsp65) and were infiltrated with activated T cells (mostly dendritic, CD3+, CD4-, CD8-, and &ggr;/&dgr; T-cell-receptor bearing). The protein was found with dendritic T cells, with immunogold-labelled hsp65 beside the dendritic processes. Thus, the immune reaction after acute arterial injury may be associated with binding and recognition of in-situ hsp65 by dendritic &g r;/&dgr; T-cells. Because

immune

Infiltration of arterial walls by activated T lymphocytes has been reported in atherosclerotic human arteries.1,2 In addition, two heat-shock proteins (hsp), hsp60 and hsp70,’e occur in atherosclerotic plaques. Such proteins may have a role in some aspects of cellular immunity as shown by the activation of human T cells by hsp antigen epitopes from mycobacteria.5 A specific association between hsp recognition and atherogenesis is suggested by studies showing that immunisation of normocholesterolaemic rabbits with mycobacterial hsp65

SJ, McLean AE, Howard C. Anticonvulsant drugs and cancer: study in patients with severe epilepsy. Lancet 1979; ii: 458-61.

a

15 Bradford-Hill A. Fallacies and difficulties: the incidence and causes of mortality. In: A short textbook of medical statistics. London: Hodder & Stoughton, 1984: 259-64.

Jancar J. Anticonvulsant drugs and cancer. Lancet 1980; i: 484. A, Kaste M, Nikkila EA, Tolppanen EM. Mortality from ischaemic heart disease among patients using anticonvulsive drugs: a case-control study. BMJ 1985; 291: 1481-83. 18 Munson JF. Death in epilepsy. Med Record 1910; 77: 58-62. 19 Moore MT. Pulmonary changes in hydantoin therapy. JAMA 1959; 16

17 Muuronen

171: 1328-33. 20

Barraclough BM.

The suicide

rate

of epilepsy. Acta Psychiatr Scand

1987; 76: 339-45. 21 Jick SS, Cole TB, Mesher RA, Tennis P, Jick H. Sudden unexplained death in young persons with Safety 1992; 1: 59-64.

primary epilepsy. Pharmacoepidemiol Drug

induces atherosclerosis.6 Additionally, antibodies to hsp65 have been detected in human carotid arteriosclerosis.7 The fact that mycobacterial hsp65 activates human T lymphocytes bearing ’y/8 T-cell receptors (TCR)8 raises the possibility that the immune mechanism in atherogenesis may be initiated by activation of this T-cell subset by hsp released after arterial injury. We studied 35 branches of the gastroduodenal, superior, and inferior mesenteric arteries resected from 8 men (ages 55-73) during gastric or colonic surgery. 22 arteries had had ligatures applied 30 minutes to 4 hours before resection; 13 non-ligated vessels served as controls. Arterial segments under and immediately distal to the ligature were divided in two: one part for electronmicroscopy and one for immunocytochemical evaluation with a series of monoclonal antibodies (details from MKH). Monoclonal antibodies to mycobacterial hsp65 (mc5205/IIH9) were given by Dr Tom Gillis (UNPD/World Bank/WHO Special Program for Research and Training in Tropical Diseases, CDCP, Atlanta, Georgia). These antibodies are specific for Mycobacterium leprae but cross-reactive for M tuberculosis. Immunoelectronmicroscopy was done with streptavidinlabelled colloidal gold particles and a biotinylated second antibody in post-fixed specimens processed for electronmicroscopy. Briefly, ultrathin sections fixed in 2-5% glutaraldehyde, post-fixed with osmium tetroxide, and stained with tannic acid were incubated sequentially with IIH9 diluted 1 in 20, biotinylated anti-mouse IgG, and 5 nm colloidal gold particles bound to streptavidin (dilution 1 in 5). To compare the difference between the immune cellular response of ligated and control arteries for T-cell count and frequency of their interactions with macrophages and hsp65, a Poisson distribution was used. Since T cells were not found in

921