The results of our national data-linkage study show the clinical trajectory of the different presentations of acute myocardial infarction in England. 49% of all people with fatal acute myocardial infarction were in hospital at the time of their death or in the 28 days preceding it and the other 51% of people who died of acute myocardial infarction deaths did not have a recent admission. Acute myocardial infarction was not mentioned as a diagnosis in 33% of deaths with a recent admission (ie, 16% of all acute myocardial infarction deaths). In about 12% of all patients admitted to hospital with acute myocardial infarction, acute myocardial infarction was diagnosed in the second or later physician encounter; it was recorded only as a comorbidity in a further 19%.
Our results are based on a large dataset that covers almost the entire population of England, where admissions to private hospitals represent only a very small proportion of total admissions, especially for acute disorders such as acute myocardial infarction. Population-based record linkage can simultaneously capture non-fatal acute myocardial infarctions, and both in and out of hospital fatal ones, allowing the clinical trajectory of the disorder to be fully followed up. Studies in a few countries where record linkage is feasible have produced estimates of acute myocardial infarction event rates, case-fatality rates, and proportion of out-of-hospital deaths,17, 22, 23, 24, 25, 26, 27 as well as other measures of quality of care.21, 28, 29 Our data have the additional advantage of allowing us to analyse the role of acute myocardial infarction that is diagnosed in the second or later physician encounter or only as a comorbidity, which have been excluded from previous analyses in England.
The use of routine data, although necessary for national surveillance, also leads to some limitations. Clinical criteria defined for the case-ascertainment of acute myocardial infarction in cohort studies do not translate well to routine ICD10 codes. Furthermore, ICD codes provide a standardised summary of a diagnosis but do not specify the underlying investigations and laboratory results—eg, ST elevation acute myocardial infarction cannot be distinguished from non-ST elevation acute myocardial infarction in ICD coding. We were thus unable to verify acute myocardial infarction events with electrocardiogram and laboratory findings, and we were reliant on the routine coding of hospital episode statistics and mortality data. The quality of hospital data varies by disorder, with common diagnoses such as acute myocardial infarction being better coded than rarer ones.30, 31 Results from linkage studies in Scotland and Wales, which used data similar to ours, show high accuracy for the diagnosis of acute myocardial infarction.32, 33 Furthermore, in two systematic reviews that examined studies comparing disease registers to routinely coded acute myocardial infarction, sensitivity and positive predictive values were 79–95%, with hospital data seeming to have better validity than death records.13, 14 For deaths, it is possible that certifying doctors use acute myocardial infarction as a convenient catch-all terminal cause when they have insufficient information about the deceased’s recent medical history. Evidence suggests that acute myocardial infarction might be over-assigned as a cause of death in some subgroups and under-assigned in others.14, 34 Guided by our results on subgroups with admissions for causes other than acute myocardial infarction preceding death from acute myocardial infarction, in-depth studies are needed to verify whether these admissions represent missed opportunities for the diagnosis and treatment of acute myocardial infarction versus misclassification of cause of death. Recorded cases aside, some studies have estimated that routinely collected hospital and mortality data might underestimate the overall incidence of acute myocardial infarction by 10–23%.35, 36 Finally, linkage itself is an imperfect process and record linkage in England fails to match 2·3% of deaths to their preceding admission.37
Our findings have three main implications. First, our results emphasise the under-investigated role of fatal acute myocardial infarction events with a recent or even current hospital admission with no mention of acute myocardial infarction or a mention only as a comorbidity. These patients make up an equal share to the mortality burden of acute myocardial infarction within 28 days of an admission with a diagnosis of acute myocardial infarction. Their hospital records often contain primary diagnoses of circulatory disorders that share risk factors with acute myocardial infarction, or mention non-specific chest pain, dyspnoea, or syncope, which might herald the impending death from acute myocardial infarction.38 The case-fatality rate for acute myocardial infarction diagnosed as a primary disorder is already relatively low, so further improvements in hospital management might only produce small reductions in mortality burden. Hospital protocols aimed at the management of acute myocardial infarction might not benefit the clinical trajectory of patients with circulatory disorders or haemodynamic instability for whom no acute myocardial infarction was diagnosed during admission. If the available clinical information can be used to prudently generate a high clinical index of suspicion, followed by risk stratification and early management, these admissions could represent opportunities to reduce the mortality burden of acute myocardial infarction.
Second, about 51% of all mortality from acute myocardial infarction is attributable to out-of-hospital deaths without a recent hospital admission. The length of delay between symptom onset and a call for help has changed little since the 1980s.39, 40, 41 More than half of these patients have symptoms (commonly chest pain, breathlessness, or a general feeling of unwellness) for a duration of more than 15 min before collapse.42 Strategies to shorten pre-hospital delays could result in substantial improvements in acute myocardial infarction survival in this group.43
Third, patients with comorbid diagnoses of acute myocardial infarction constitute roughly 19% of all admissions related to acute myocardial infarction and have case-fatality rates two to three times higher than those of patients with a main diagnosis of acute myocardial infarction. Patients with comorbid acute myocardial infarction diagnoses tend to be older and are more likely to be women than patients with main diagnoses and often have a stressor condition such as hip fracture or pneumonia recorded as the primary reason for admission. The universal definition of myocardial infarction subclassifies acute myocardial infarction into type 1 acute myocardial infarction, in which cardiac ischaemia is caused by the formation of an acute thrombus in the coronary arteries, and type 2, in which it is caused by an imbalance in the supply and demand for myocardial oxygen.44 Although we cannot be certain without full record review, the timing and secondary prioritisation of most acute myocardial infarctions recorded as comorbid events would suggest that they belong to the type 2 category, especially those accompanying disorders such as atrial fibrillation. The optimal management of type 2 acute myocardial infarction is currently uncertain, with risk assessment and optimisation of haemodynamic status remaining the mainstay of treatment for these patients.
The underlying method to reduce mortality after acute myocardial infarction is timely contact with the health system and diagnosis of the acute myocardial infarction. Our findings on the distributions of deaths from acute myocardial infarction show that substantial reductions in acute myocardial infarction mortality will require attention to the large proportion of these deaths that are not preceded by a hospital admission or are preceded by an admission for a another cause.
Article Info
