Introduction

Twenty five years ago at least two thirds of deaths from acute coronary heart disease happened outside hospital,^1-3 and data from the monitoring trends and determinants in cardiovascular disease (MONICA) study showed that this was still the case in many countries during the 1980s.⁴ Mortality from coronary heart disease has, however, declined by about 30% in England and Wales since 1980.⁵ Moreover, devolution of prehospital care to paramedical staff, public training in cardiopulmonary resuscitation, and efforts to reduce delay in giving thrombolytic treatment should all ensure that the benefits of therapeutic advances are more readily available to victims of acute heart attack.

The aim of the United Kingdom heart attack study is to re-examine the incidence, treatment, and outcome of acute heart attacks by studying events in three health districts in England and Wales during 1994-5. This report describes total case fatality, with particular reference to events occurring outside hospital and the impact of resuscitation attempts outside hospital. The outcome of myocardial infarction in patients admitted to hospital is reported elsewhere.⁶

	Subjects and methods

All deaths from acute episodes of ischaemic heart disease and all cases of acute myocardial infarction in hospital in people under 75 years of age were recorded over two years (January 1994 to December 1995) in the three health districts of Brighton, South Glamorgan, and York. Methods were established during a pilot study carried out in Brighton during 1993 ^{7 8} and, in the light of the experience gained, were agreed on before the main study was started.

Definition of population
To avoid bias from recording deaths outside and inside hospital in different populations, we selected our catchment population to include only the areas served by the study hospitals (one each in Brighton and York and two in South Glamorgan); patients were classed by postcode. Population figures broken down by age and sex were supplied by the Office for National Statistics. Defined in this way, the population of Brighton (Brighton, Hove, and Lewes) was 282 000, that of South Glamorgan (Cardiff and the Vale of Glamorgan) 408 000, and that of York (city and surrounding, mainly rural, area) 264 000. The total study population (all ages) was 954 000. People whose events occurred within the study area who were visitors to the area (275, or 8% of total cases) were recorded as non-resident but were included on the assumption that similar numbers of events occurred in visitors as in residents who had gone outside the area.

Selection and identification of cases
Deaths outside hospital were included if ischaemic heart disease was the principal cause of death at coroners' necropsy (86% (959/1114)) or, when there was no necropsy, if the patient had a history of ischaemic heart disease and had died suddenly or after prolonged chest pain and had no other apparent cause of death (14% (155/1114)). Cases at necropsy were required to have at least 50% stenosis of the diameter of one major epicardial coronary artery, with or without recent coronary thrombus or old or recent myocardial infarction. Deaths from chronic heart failure due to ischaemic heart disease were recorded in one centre (Brighton) but were not included in this analysis. They comprised 11% of all coronary deaths in Brighton in the age range studied, and only 29% occurred outside hospital.

View larger version (24K): [in this window] [in a new window]   Fig 1.    Case fatality and proportion of fatal events occurring outside hospital in the three centres. Circles indicate mean values and bars 95% confidence intervals

Case fatality and statistics
Case fatality rates were calculated as the total number of deaths divided by the total number of events as defined and were expressed as percentages. Events occurring more than 30 days after a previous non-fatal episode were considered to be new events. Patients who were brought into hospital after having had a cardiac arrest outside hospital who were pronounced dead or who later died in hospital were included as deaths outside hospital because the event directly leading to death had occurred before arrival at hospital.

	Results

Figure 1 shows the numbers of cases, case fatality rates, and the proportion of fatal events occurring outside hospital for the individual centres. Of the 3523 events which we recorded, 3476 occurred in white people and 934 in women; 1589 (45% (95% confidence interval 43% to 47%)) were fatal within 30 days of the event. Case fatality was lower in York than in the other centres for all age groups (42% (466/1117) v 47% (1123/2406); P=0.02) and for ages <55 years (19% (40/206) v 31% (120/389); P=0.01) but not for age groups 55-64 and 65-74 years.

Figure 2 shows the numbers of survivors and deaths at each stage of the illness. Overall, 1172 of the 1589 fatal events (74% (72% to 76%)) occurred outside hospital. Of these, 1095 occurred in people who were not admitted to hospital, 21 occurred after discharge from hospital but within 30 days of the infarction, and 56 occurred in patients who were resuscitated outside hospital but later died in hospital. The remaining 417 (26%) fatal events occurred in hospital. Forty six per cent of the patients who died (737/1589) were known to have ischaemic heart disease, and this proportion rose with age from 31% (50/160) at <55 years to 49% (503/1019) at age 65-74. 

View larger version (33K): [in this window] [in a new window]   Fig 2.    Flow diagram showing numbers of survivors and deaths at each stage

Table 1 shows the relation between the place of occurrence of the fatal event, age, and sex, and figure 3 shows the relation with age alone. Total 30 day fatality (upper line in fig 3) rose with age from 27% (23% to 31%) at age <55 to 53% (51% to 55%) at 65-74 years. The lower line in figure 3 represents fatality outside hospital. Thus the portion between the two lines represents fatality in hospital, which was very low (3% of those admitted to hospital) for patients under 55 years of age but rose to 29% for those aged 70-74 years. Corresponding figures for fatality outside hospital were 24% and 37%, so that the increase in fatality outside hospital with age was much less than the increase in fatality in hospital. Consequently, the proportion of fatal events occurring outside hospital was age dependent, falling from 91% (87% to 95%) at age <55 years to 77% (73% to 81%) at 55-64 years and 70% (67% to 73%) at 65-74 years. The (negative) gradient for age as a determinant of whether death occurred outside or inside hospital was highly significant (P<0.001) for men (table 1) and for both sexes combined (fig 3), but it was not significant for women.

View larger version (18K): [in this window] [in a new window]   Fig 3.    Total case fatality and case fatality outside hospital by age group

Premonitory symptoms, usually chest pain, during the few hours before the cardiac arrest were reported by bereaved relatives in 38% of deaths outside hospital (441/1172), while death seemed to be truly sudden in 13% (150/1172). In 49% of cases (581/1172) it was impossible to establish whether symptoms had been present, usually because the victim was found dead, having last been seen alive several hours (median 7 hours) previously. The commonest symptom was chest pain, but symptoms of "breathlessness," "indigestion," or "feeling unwell" were also reported frequently. There was a tendency for premonitory symptoms to be reported more often in younger than in older subjects, but this was not significant.

Effect of resuscitation outside hospital on case fatality
Of the 1227 cardiac arrests outside hospital, 920 occurred at home, 203 in a public place, 79 in doctors' surgeries, ambulances, or nursing homes, and 25 at work. Fifty five of the 111 people who were successfully resuscitated outside hospital recovered in hospital and survived to 30 days. Of these, 51 had documented ventricular fibrillation, pulseless ventricular tachycardia, or asystole, while 4 responded to basic life support and were thought to have had a cardiac arrest on clinical grounds. Had these 55 patients not been resuscitated, the total case fatality would have been 46.7% not 45%.

	Discussion

The most important findings in this study were the high proportion (74%) of fatal events that occurred outside hospital and the inverse relation with age, 91% of fatalities at age <55 and 70% at age 65-74 happening outside hospital. A similar age trend has been reported from the United States on the basis of data from death certificates⁹ and was also observed in the pilot Brighton heart attack study.⁷ In the Glasgow MONICA study men were more likely than women to die outside hospital¹⁰; whether the likelihood of fatal events occurring outside hospital was related to age was not stated, but patients aged >64 years are not included in MONICA studies.

What is the reason for this disturbing finding? Hospital treatment has improved greatly over the past 20 years, resulting in a fatality rate in hospital for our patients who were under 55 years of age of only 3% (fig 3). If fatality outside hospital has not improved to the same extent, this would increase the proportion of deaths outside hospital for younger patients. Again, we did not record unclassifiable deaths outside hospitalthat is, those in which death was attributed to coronary disease without results from necropsy or a history of clinical manifestation.⁴ We found such cases to be confined mainly to those aged 65-74, who were not included in the MONICA study.¹¹ Inclusion of such cases would have increased the proportion of deaths outside hospital and diminished the age gradient.

The high proportion of deaths outside hospital shows the potential limitations of further improvements in hospital treatment, particularly for younger patients. It emphasises that further large reductions in mortality can be accomplished only by primary prevention, secondary prevention, or intervention before admission.

Three potential strategies exist to improve intervention before admission: improved responses by ambulance crews, intensified training of members of the public in cardiopulmonary resuscitation, and public education on the importance of dialling 999 for prolonged chest pain. These strategies might result in more cardiac arrests being witnessed by paramedical staff and more patients having successful defibrillation.

Improved ambulance response
The most successful centres have a dual¹² or triple¹³ response system in which the immediate response to an emergency call is made by a normal ambulance or by the fire service, followed by paramedical staff trained in advanced life support who may have to travel a longer distance. A previous analysis found that dual response provides an optimum success rate,¹⁴ and success is enhanced by providing those who first attend the scene with automatic defibrillators.¹⁵ In the United Kingdom all frontline ambulances carry defibrillators, and one member of the ambulance crew has usually had paramedical training. The NHS is planned to continue with a single paramedic response system but with prioritisation of emergency calls so that response times for life threatening emergencies will be reduced from the present 14 minutes in 95% of urban areas to 8 minutes for 90% of calls in all areas.¹⁶

Citizen training in cardiopulmonary resuscitation
Basic life support was given by bystanders in 27% of witnessed arrests in our series, and it reduced fatality modestly but significantly (table 2). Interestingly, the proportions receiving bystander life support were little different at 22% in Helsinki,¹³ 18% in Gothenburg,¹⁸ and 36% in Seattle.¹⁴ In the unlikely event that the proportion receiving basic life support could be doubled in the United Kingdom, and assuming that the proportions in table 2 were replicated, the reduction in fatality would be less than 0.5%. These figures support previous opinions that present benefits from bystander cardiopulmonary resuscitation are real but limited.¹⁹ Protocols need to be rigorously taught, retaught, and remembered; failure by bystanders to dial the emergency number before doing anything else is another problem.¹⁹ The ideal place to start training may be in school.²⁰ Notwithstanding the above, the contribution of bystander life support to survival would very likely be larger if ambulance response times could be improved.¹⁶

Public education
A much more impressive result (40% (28% to 53%) survival) was seen among those of our patients who experienced cardiac arrest in the presence of paramedical staff equipped with a defibrillator. These people, comprising only 5% of those who had arrests outside hospital, were fortunate enough to have reported premonitory symptoms in time, and the ambulance responded promptly. We have shown in a pilot study that there is a marked deficiency in public knowledge about the causation of myocardial infarction and in particular the differentiation of heart attack from cardiac arrest.²¹ Many, and probably most, people who died outside hospital in our series had premonitory symptoms, usually chest pain, often lasting for several hours. Similar findings have been reported recently from Glasgow.²² Although the success of media campaigns to induce patients with prolonged chest pain to seek help early has been limited,²³ we believe that provision of information to the general public so that they can make informed decisions about the action to take for suspected heart attack can only be beneficial. Fresh initiatives in public education are necessary. A useful message is to dial an emergency number (999) for chest pain lasting 15 minutes or more.

Prevention of death outside hospital
Primary prevention of coronary heart disease is the ideal, and reduction in coronary risk factors has been a major contributory factor to the declining mortality from coronary heart disease.²⁴ A recent analysis suggests, however, that more has been achieved and is still achievable by secondary than by primary prevention.²⁵ An important finding from our study was that 46% of the patients who died were already known to have ischaemic heart disease. Recent advances in secondary prevention, particularly in lowering cholesterol concentration with drugs that inhibit 3-hydroxy-3-methylglutaryl coenzyme A reductase, ^{26 27} have shown a 20-30% reduction in total deaths in patients with a history of angina or previous myocardial infarction. A reduction in mortality by 25% among those in our study who were known to have coronary heart disease and who died from events occurring outside hospital would have saved about 183 further lives. Additional benefits might also have been obtained from more strenuous advice to stop smoking and wider use of antiplatelet drugs and  blockers, none of which seemed to have been used to their full potential in subjects known to have coronary disease (data not shown).