Inpatient deaths from acute myocardial infarction, 1982-92: analysis of data in the Nottingham heart attack registerBMJ 1997; 315 doi: https://doi.org/10.1136/bmj.315.7101.159 (Published 19 July 1997) Cite this as: BMJ 1997;315:159
- Nigel Brown, research fellow in cardiologya,
- Tracey Young, research assistantb,
- David Gray, reader in medicinea,
- Allan M Skene, directorb,
- John R Hampton, professor of cardiologya
- a Division of Cardiovascular Medicine, University Hospital, Queen's Medical Centre, Nottingham NG7 2UH
- b British Heart Foundation Cardiovascular Statistics Unit, Department of Mathematics, University of Nottingham, Nottingham
- Correspondence to: Dr Brown
- Accepted 15 April 1997
Objective: To assess longitudinal trends in admissions, management, and inpatient mortality from acute myocardial infarction over 10 years.
Design: Retrospective analysis based on the Nottingham heart attack register.
Setting: Two district general hospitals serving a defined urban and rural population.
Subjects: All patients admitted with a confirmed acute myocardial infarction during 1982-4 and 1989-92 (excluding 1991, when data were not collected).
Main outcome measures: Numbers of patients, background characteristics, time from onset of symptoms to admission, ward of admission, treatment, and inpatient mortality.
Results: Admissions with acute myocardial infarction increased from 719 cases in 1982 to 960 in 1992. The mean age increased from 62.1 years to 66.6 years (P<0.001), the duration of stay fell from 8.7 days to 7.2 days (P<0.001), and the proportion of patients aged 75 years and over admitted to a coronary care unit increased significantly from 29.1% to 61.2%. A higher proportion of patients were admitted to hospital within 6 hours of onset of their symptoms in 1989-92 than in 1982-4, but 15% were still admitted after the time window for thrombolysis. Use of ß blockers increased threefold between 1982 and 1992, aspirin was used in over 70% of patients after 1989, and thrombolytic use increased 1.3-fold between 1989 and 1992. Age and sex adjusted odds ratios for inpatient mortality remained unchanged over the study period.
Conclusions: Despite an increasing uptake of the “proved” treatments, inpatient mortality from myocardial infarction did not change between 1982 and 1992.
During 1982-92 major changes in management of myocardial infarction in an unselected population have been guided by the results of randomised trials
Adjusted odds ratios for deaths in hospital from acute myocardial infarction did not change over this period despite an overall fall in recorded deaths from ischaemic heart disease in Nottingham
The use of existing treatments needs to be optimised and new management strategies need to be introduced if inpatient mortality from myocardial infarction is to be reduced
Coronary care units were first established in the 1960s,1 and subsequently inpatient mortality fell from 23-40% to 16-18%.1 2 3 It was generally believed that the fall was secondary to the detection and appropriate management of serious arrhythmias. More recently, large multicentre randomised trials have documented improvements in outcome after the use of aspirin,4 ß blockade,5 and thrombolysis,4 6 7 all of which have become standard management in myocardial infarction. In an overview of the thrombolytic trials, mortality at 35 days was 9.6% in those treated with thrombolysis compared with 11.5% in controls,8 although the numerous exclusion criteria pertaining to trials mean that the general population of patients with infarction may not experience such an improvement. Although most deaths from myocardial infarction occur in the first two hours after the onset of symptoms and outside hospital, the above clinical trials have been concerned with hospital based treatments that are effective at improving survival.
We documented from a hospital perspective all admissions of patients with myocardial infarction in a defined community over 10 years. We determined the uptake of proved treatments and observed whether any changes in inpatient mortality might be related to the previously documented fall in overall death rates from ischaemic heart disease in Nottingham.9
The methods of data collection for the Nottingham heart attack register have been previously described,10 but in brief all patients admitted to Nottingham's hospitals with symptoms suggesting acute myocardial infarction were identified prospectively, and an extensive record of management and outcome was documented. Patients were assigned to the following diagnostic categories after the hospital case records had been reviewed by the physician responsible for the register at the time: definite myocardial infarction; possible myocardial infarction; ischaemic heart disease; chest pain, unknown cause; and “other diagnosis.” For this study we retrieved data for analysis for the years before (1982, 1983, and 1984) and after (1989, 1990, and 1992) the widespread introduction of aspirin, ß blockers, and thrombolysis (data were not collected in 1991). We included data only on patients who had had a definite myocardial infarction—namely, a convincing history plus either changes in the electrocardiogram that were diagnostic or a rise in cardiac enzymes to more than twice the upper limit of normal, or both. Changes over time in patients with a possible infarction will be reported separately.
The methods of data collection and the diagnostic criteria10 used have remained the same since the register was established despite inevitable changes of staff over the years. Detailed comparative tests were undertaken during changeover of staff to ensure consistency of diagnosis. Patients who had a cardiac arrest outside hospital and who never recovered consciousness were not included in this analysis.
We recorded age, sex, length of stay, “new” pathological Q wave or non-Q wave myocardial infarction, site of infarction, and electrocardiographic evidence of previous infarction. For 1989 onwards we collected information on documented history of previous myocardial infarction, previous coronary revascularisation, and Killip class (clinical estimate of infarct severity)11 on admission. We documented time of onset of symptoms to admission to hospital for those who could recall either accurate times or a narrow time band—for example, three to four hours—and admission or transfer to a coronary care unit within 24 hours or management on a general medical ward. We noted details of inpatient treatment with antiarrhythmics, ß blockers, anticoagulants (heparin, warfarin), digoxin, diuretics, angiotensin converting enzyme inhibitors, aspirin, and thrombolysis (streptokinase or alteplase) and recorded outcome in hospital.
Changes in discrete variables over time were analysed by using the χ2 test. Linear models were fitted to determine whether trends had occurred over time for mean age and mean length of admission, the error terms for continuous variables being checked for normality first. A log transformation was performed on length of stay in hospital. All tests of significance were two tailed, with P<0.05 considered to be significant.
Crude odds ratios for mortality were calculated by using 1982 as the standard year (odds ratio=1) and then adjusted in a logistic regression model to take into account changes in age and sex across the years. The S-PLUS statistical package was used for all analyses.12
Demographic and clinical characteristics
The total number of patients admitted with a suspected myocardial infarction rose from 2042 in 1982 to 4717 in 1992. The number with a final diagnosis of definite myocardial infarction, however, rose much less, from 719 in 1982 to 960 in 1992. The largest increase was in patients with possible infarction, previous evidence of ischaemic heart disease, or chest pain with an unknown cause.
Table 1) shows that the increase in the number of patients admitted with myocardial infarction was 70% in those aged 70 to 74 years and 200% in those aged 75 years and over.
Table 2 shows that between 1982 and 1992 the mean age for both men and women admitted with acute myocardial infarction increased significantly, and the proportion of male patients admitted fell. The mean length of stay fell by 1.5 days; the proportion of patients with only electrocardiographic evidence of previous infarction (old Q waves) fell, as did, to a lesser degree, the proportion of those with a definite previous myocardial infarction (from medical history or electrocardiogram). There was no change in the small numbers of patients with previous coronary revascularisation. The proportion of patients with a new Q wave infarction increased significantly in 1992, but the proportion of patients with an anterior myocardial infarct did not change. Killip class was recorded only from 1989 onwards, but between 1989 and 1992 there was an increase in the proportion of patients in class 1, a small increase in class 3, a fall in class 2, and no change in class 4 (cardiogenic shock).
Admissions to coronary care
Table 3 shows the increasing numbers of patients admitted to coronary care. (Overall, about 75% of all patients with infarction admitted to hospital in Nottingham are cared for in such units.) The proportion of those aged 75 years and over who were managed in the unit rather than an ordinary medical ward doubled. However, the proportions of patients aged under 55 years and 55 to 64 years admitted to coronary care between 1982 and 1992 fell from 96.6% to 77.9% and from 94.2% to 77.0% respectively—equivalent to almost two young patients being managed on the general ward each week. Over the 10 years the proportion of women admitted to coronary care increased significantly, from 60% in 1982 to 67% in 1992 (χ2=23.4, df=5, P<0.001). Men were still more likely than women, however, to be admitted to the unit in 1992 (χ2=11.28, df =1, P<0.001).
Time from onset of symptoms to admission
A mean of 86.7% of patients over the study period were able to provide timing data, and table 4 shows the time from onset of symptoms to admission. The proportion of patients admitted within two to six hours of onset of their symptoms improved significantly between 1982-4 and 1989-92 (P<0.001); the proportions of those admitted within 1 hour, 7-12 hours, and 13-24 hours, however, hardly changed. Fewer patients were admitted more than 24 hours after onset of symptoms in 1989-92.
Management in hospital
Table 5) details temporal changes in management and shows a significant fall in the use of antiarrhythmics, a threefold increase in the use of ß blockers over 10 years, and a twofold increase in the use of angiotensin converting enzyme inhibitors between 1989 and 1992. There was a high uptake of aspirin use after its widespread introduction in 1988 and a 1.3-fold increase in the use of thrombolysis from 1989 to 1992. The use of diuretics and digoxin declined during the years of the study.
The proportion of patients with contraindications to ß blockers (owing to existing medical conditions) varied from 3% to 10%, to aspirin from 1% to 2%, and to thrombolysis from 6% to 15%.
Overall inpatient mortality from myocardial infarction rose from 16.1% to 21.7% between 1982 and 1992 (table 6). However, age and sex specific inpatient mortality did not significantly change in any age or sex group over the 10 years.
The adjusted odds ratios for death from myocardial infarction, with allowance for the effects of age and sex, showed no significant change in mortality over 10 years (table 6).
Our data clearly show an increase in admissions with acute myocardial infarction over the 10 years, although this increase is mainly confined to patients aged 75 years and over, and to a lesser extent to those aged 70 and over. We believe that the increase is partly a reflection of the increasingly elderly population13; partly a greater awareness by patients of the significance of symptoms of chest pain; and partly a lower threshold for admission in accident and emergency departments and greater surveillance by medical staff. Although we have previously shown that the treatment of patients with myocardial infarction at home was relatively uncommon in the early 1980s,14 we believe that this is even more unusual now with the known benefits of thrombolysis for patients of all ages. Rather than the incidence of myocardial infarction increasing, it seems that more people are now admitted with suspected infarction so that detection rates of definite myocardial infarction are greater. A similar increase in total admissions with myocardial infarction, predominantly elderly people, has been documented elsewhere.15
Women and older patients, despite a worse prognosis, have historically been less likely to be admitted to a coronary care unit.16 17 There has been an encouraging increase in admissions of these groups, in particular elderly people, to the Nottingham units such that over 60% of the patients aged over 75 were admitted to coronary care in 1992.
Mortality in patients with recurrent myocardial infarction is twice that of counterparts with a first event18 19. Analysis of background characteristics showed a fall in the proportion of patients with only electrocardiographic evidence of previous infarction and, to a lesser extent, in those with a documented history of myocardial infarction, which might be expected to improve inpatient outcome. A reduction in duration of hospital stay over the 10 years might additionally be expected to lower mortality artefactually, but the effects of this are probably minimal as most deaths occur within the first two days of admission.4 More patients in 1992 had evidence of a Q wave infarction, but opinions conflict about the relevance of this to outcome,20 21 22 23 24 with little sensitivity or specificity of Q waves with respect to true transmural infarction.25 26
Despite some improvement, considerable delays still occur between the onset of symptoms and admission, and in 1992 three in 20 patients were admitted after the 12 hour time window for thrombolysis irrespective of any delays in hospital. For a “typical” year in Nottingham this equates to over 100 patients. The relation between the delay in admission and the outcome is not a simple one. We have reported that confounding by the lack of data on times of onset of symptoms in those who die soon after admission makes it difficult to assess the relation of delays to mortality.14 Although earlier admission could lead to a paradoxical increase in mortality (as a result of admission of patients who might have died in the community), our baseline data do not suggest that we are dealing with more severe infarctions in the later years of the study. The proportion of patients with previous Q wave infarction fell; the proportion with anterior infarction remained the same; the proportion given diuretics (a marker of those with clinical cardiac failure) fell slightly; and, although data on Killip class are available only for the later years, the proportion of patients with cardiogenic shock has not changed. Other research has similarly not shown significant changes in severity of infarction over time.27 28 Other factors, however, such as comorbidity, for which we cannot control might have masked an improvement in survival.
The use of prophylactic antiarrhythmics fell after lack of evidence of benefit in treating warning arrhythmias and the results of the cardiac arrhythmia suppression trial.29 The use of ß blockers increased, although, even after allowance for those with contraindications, 40% of patients in 1992 might still have benefited from this treatment but did not receive it. Although some patients may start receiving the drug in outpatient departments, the poor uptake of ß blockers is not unique to Nottingham.30 31 The use of diuretics and digoxin fell slightly over the 10 years, and, although the proportion of patients receiving these drugs in 1992 seems high, our findings are similar to another recent report in unselected patients with myocardial infarction.32 The increasing use of anticoagulants is the result of the use of intravenous and subcutaneous heparin as concomitant treatment in several thrombolytic trials during the years 1989 to 1992. Only 6.4% of patients were taking oral anticoagulants at discharge in the years 1989 to 1992.
Thrombolytic treatment was not used in 1982-4, but by 1992, 48% of patients received this treatment. Estimates of the proportion of all patients with myocardial infarction likely to be suitable for treatment vary from 25% to 33% in the United States31 to between 70% and 80% in Britain,33 although we believe the figure for Britain to be optimistic. Our figures suggest that 15% of patients may be admitted outside the time window of 12 hours, 15% may have a contraindication, and not all will have or develop electrocardiographic criteria for thrombolysis. Of our cohort of patients in 1992 who did not receive thrombolysis, had no contraindication, and were admitted in under six hours (the general policy which preceded publication of the late assessment of thrombolytic efficacy study34), 73 (5%) subsequently developed electrocardiographic criteria for lytic treatment8 but for unknown reasons did not receive it.
Disappointingly we have seen little change in inpatient mortality over the 10 years of our study. Our findings highlight the differences between the selected patients of clinical trials and the general population of patients, who have an overall mortality of 20%, at least twice that of most patients in trials. Purchasers and those involved in clinical audit need to be aware that the proportional reductions in mortality seen in clinical trials do not necessarily translate into benefits in a general population.
We have previously shown that overall mortality from ischaemic heart disease is falling in Nottingham,9 which may be due to changes in the natural course of the disease, a reduction in community cardiac risk factors, and improved treatment of chronic ischaemic heart disease, as suggested elsewhere.35 36 37 This experience is not unique to Nottingham. Goldman and colleagues in the United States in the 1970s similarly found that, although overall mortality from ischaemic heart disease had fallen, inpatient mortality from myocardial infarction was static38; however, this was in the years before thrombolysis and treatment with antiplatelets and ß blockers. A report from the United States showed that overall inpatient mortality from myocardial infarction rose from 13.1% in 1984 to 16.8% in 1988, with no clear trends in age and multivariate adjusted mortality.39
Dellborg and colleagues from Sweden, however, showed a reduction in inpatient mortality between 1979 and 1990, even in an elderly population, on the basis of data from a register for a coronary care unit.40 Further studies in Ontario15 and the United Kingdom41 have shown reductions in overall mortality over similar time periods of 21.0% to 17.1% and 25.4% to 20.2% respectively, but both these studies were based on hospital discharge codes and susceptible to the errors associated with these.
Major management changes have occurred over the period of our study as the lessons of controlled randomised clinical trials have been applied. We have shown that, despite some improvement, patients still delay seeking help for a considerable time. Undoubtedly current treatment in management of myocardial infarction, particularly thrombolysis and ß blockade, needs to be optimised, and new strategies need to be introduced. A reduction in our inpatient mortality from acute myocardial infarction remains elusive.
We are grateful to the staff of the register for their concerted efforts over the years and acknowledge the contribution of Dr John Rowley, who collected data during 1982-4 for the Nottingham heart attack register.