Patients with ST-elevation myocardial infarction (STEMI) who die in hospital before inpatient admission are generally not included in clinical studies and registries, and the clinical profiles of patients who die earlier versus later are not well defined. We aimed to characterize all patients with STEMI who arrived at emergency departments in the province of Quebec (Canada) based on inpatient admission status and when they died. All patients who presented with symptoms and core laboratory–confirmed STEMI or left bundle branch block during 6 months in 82 hospitals in Quebec were included. Death certificates were used to identify nonadmitted deaths. Of the 2017 patients with STEMI, 340 (16.9%) died within 1 year. Of the latter, 63 (18.5%) were nonadmitted deaths (group A), 179 (52.6%) were deaths after admission but within 30 days (group B), and 98 (28.8%) were deaths after 30 days to 1 year (group C). Group A was younger and most often hemodynamically unstable, followed for both features by B then C. Earliest presentation from symptom onset and most frequent ambulance use were found in group A, followed by B, then C. Presenting electrocardiogram (ECG) features were most severe in A, then B, then C (more arrhythmias, more anterior STEMI, more leads with ST elevation, and higher ST elevation). Patients who died earliest had the least frequency of previous myocardial infarction, coronary revascularization, vascular disease, and heart failure, and the least noncardiac co-morbidity. In conclusion, patients with STEMI dying in hospital before inpatient admission contributed substantially to overall STEMI mortality. Although dying patients who presented earlier had severer presenting clinical profiles, they were paradoxically younger and had less co-morbidity. Previous co-morbidities may favor adaptive protective mechanisms on initial presentation with STEMI.
The scientific reports that address ST-elevation myocardial infarction (STEMI) and its clinical presentation, management, and treatment are essentially based on randomized clinical trials, registries, and medicoadministrative data. However, patients in randomized trials tend not only to be highly selected but must also be alive when they are considered for enrollment and when subsequently recruited. Registries and medicoadministrative databases generally glean information from patients who have been admitted (i.e., have formal inpatient status). Thus, patients who present to emergency departments (EDs) with STEMI, but who die before they obtain inpatient status, will generally not be found in these data collections (for example USA’s ACTION Registry-GWTG for hospitalized patients ). Although some information about early deaths due to STEMI may be found within ED and prehospital databases, quantification and characterization of these patients in relation to patients with STEMI who die later and those who survive have not been previously performed. Such inquiry may provide novel information about the dynamic determinants of mortality and survival in patients with STEMI.
The objective of this study was to identify, characterize, and compare patients arriving at hospital EDs with STEMI who die during different time intervals, including those who die before inpatient admission and are thus “under the radar” of most clinical investigations (we refer to them as “nonadmitted” patients). This study was conducted as part of a province-wide field evaluation of STEMI.
Methods
During a 6-month period (October 1, 2008, to March 31, 2009), we carried out a systematic second field evaluation of STEMI care in the province of Quebec, Canada (population 8.1 million) at the request of the provincial Ministry of Health. Methods and results of our first field evaluation (2006 to 2007) have been previously published. A total of 82 hospitals, that treated >95% of all documented myocardial infarctions in Quebec, participated in this evaluation whose objective was to evaluate treatments, delays, and clinical outcomes. Briefly, a designated medical record librarian in each hospital, in collaboration with a librarian on the research team, examined consecutive charts of all patients admitted with myocardial infarction in the evaluation period. Patients with STEMI were identified using an automated algorithm, and the relevant chart review data were extracted and entered into a secure, centralized Web site. Two cardiologists at the coordinating center each retrospectively examined all the first inhospital electrocardiograms (ECG), blinded to all patient clinical and process-of-care features. We identified all admitted patients with presenting symptoms suggestive of acute myocardial infarction, a final discharge summary diagnosis of acute myocardial infarction, and ECG core laboratory–confirmed STEMI or left bundle branch block (LBBB) according to at least 1 study cardiologist. For all patients, we linked chart review data to prehospital (ambulance) databases and medicoadministrative databases, as previously described.
To identify patients who had STEMI and who died in hospital without inpatient admission, we reviewed all death certificates in all participating hospitals during the same 6-month period and retrieved patient data from the matching ED records. The nonadmitted patients eligible for the present study had to have all the following: (1) mention of acute myocardial infarction in the medical chart and/or death certificate; (2) documented ED arrival with suggestive symptoms or so specified in prehospital notes; (3) a technically interpretable ECG; and (4) ECG core laboratory confirmation of STEMI or LBBB. We compared the clinical profiles, including the Thrombolysis In Myocardial Infarction (TIMI) risk index, of the patients who died before hospital admission (group A) with admitted patients who died within 30 days of admission (group B), who died beyond 30 days to 1 year (group C), and a random sample of 300 admitted patients with STEMI from the field evaluation who were alive at 1 year (group D). Information on co-morbidities preceding the index STEMI event was obtained by linkage to Quebec’s hospital discharge database, going back 5 years for all patients. Deaths after hospital discharge for admitted patients were ascertained by linkage to the Quebec death registry. This study was undertaken by the Cardiovascular Evaluation Unit of INESSS ( Institut national d’excellence en santé et en services sociaux ), a publicly funded Quebec health services and technology evaluation organization. Study approval was obtained from each of the 82 hospitals and from the Quebec Provincial Commission for Access to Information that waived the need for patient consent as neither intervention nor patient contact was involved.
The sample of 300 admitted patients who were alive at 1 year was selected using a computerized random number generator and weighted by regional representation. Median, twenty-fifth percentile, and seventy-fifth percentile times to death were calculated. Patients in groups A through D were compared with respect to age, gender, clinical characteristics at presentation, characteristics on the first ED ECG, co-morbidities documented in the previous 5 years (as separate conditions and as the composite median Charlson index ), and process-of-care characteristics for the index STEMI event (means of transport to hospital [ambulance vs other], delays from the first medical contact, reperfusion treatment). The differences among groups A, B, and C were tested for statistical significance using analysis of variance for means, the Kruskal–Wallis test for medians, and the chi-square test for 2 × 3 tables (or Fisher’s exact for sparse data). When statistically significant differences between the 3 groups were found (p <0.05), a multiple comparison test was performed to confirm pairwise differences (Tukey for means, Kruskal–Wallis for medians, and chi-square/Fisher’s exact for proportions); the Bonferroni correction was also applied when necessary (i.e., for tests of medians and proportions). All statistical analyses were performed using SAS version 9.3.
Results
During the study period, we identified 2,017 patients who met the eligibility criteria; of these, 340 (17%) died within 1 year. Of the latter, 63 (18%) were in group A (nonadmitted deaths), 179 (53%) were in group B (admitted deaths up to 30 days; 123 of these occurred as inpatients), and 98 patients (29%) were in group C (admitted deaths between 31 days and 1 year; 5 of these occurred as inpatients). Thus, group A, the nonadmitted patients who died and who would usually not be counted in reporting outcomes, comprised nearly 1/5 of all deaths at 1 year and raised 1-year mortality from 14% (if not counted) to 17%, a 21% relative increase. The contribution of group A to STEMI mortality was even greater if only shorter-term deaths were examined; they accounted for 33% of overall inhospital (ED or inpatient) STEMI deaths, raising inhospital mortality from 7% (excluding group A) to 10%, a 43% relative increase. Group A comprised 26% of all 30-day deaths and raised 30-day STEMI mortality from 9% to 12%, a 33% relative increase. Median time from triage to death was 55 minutes (twenty-fifth to seventy-fifth percentiles: 29 to 146 minutes) for group A patients versus 3 days (1 to 9) for group B and 120 days (63 to 230) for group C.
Clinical characteristics of the patients who died within the 3 time intervals and of the admitted group that survived (group D) are provided in Table 1 . Although survivors were youngest and were most often men, group A patients who died were significantly younger and tended to be more often male than admitted patients who died. A low or high heart rate on presentation was found least often in survivors and most often in group A. There was a striking gradient in the proportion of patients who were hypotensive on presentation: highest in group A, second highest in group B, then group C, and least often in survivors. Accordingly, the same gradient was found for the TIMI risk index. Median symptom duration to ED triage was less than an hour for group A and 2 hours or slightly more for groups B and C.
| Variable | Group A | Group B | Group C | p value ∗ | Group D |
|---|---|---|---|---|---|
| Non-admitted deaths (n=63) | Admitted 30-day deaths (n=179) | Admitted 31-day to 1-year deaths (n=98) | Admitted survivors at 1 year (n=300) | ||
| Mean age (years) ± standard deviation | 71 ± 15 | 76 ± 13 | 78 ± 11 | 0.04 | 61 ± 13 |
| Women | 20 (32%) | 76 (42%) | 49 (50%) | 0.07 | 73 (24%) |
| Heart rate (bpm) <60 or >100 | 32 (52%) | 92 (51%) | 39 (40%) | 0.14 | 94 (31%) |
| Systolic blood pressure (mmHg) <90 | 34 (55%) | 43 (24%) | 9 (9%) | <0.0001 | 7 (2%) |
| Median TIMI† index (interquartile range) | 61 (36-91) | 47 (32-62) | 43 (30-58) | 0.002 | 19 (14-28) |
| Median symptom duration to triage (min) (interquartile range) | 50 (35-74) | 123 (63-244) | 132 (61-344) | <0.0001 | 102 (60-172) |
ECG characteristics are listed in Table 2 . LBBB was found far less frequently in survivors. Groups A and B had less LBBB than group C. However, right bundle branch block was most common in group A, followed by B, then C that resembled D. Excluding LBBB, ECG signs of STEMI severity such as maximum ST-elevation, the number of leads with ST-elevation, and the proportion with QRS conduction delay (≥110 ms) followed this same gradient: most intense in group A, followed by group B, then group C, the latter resembling the survivor group. The highest proportion of anterior STEMI was found in group A. Among those who died, group A had the lowest proportion of patients with Q-waves already present. Arrhythmias were most frequent in group A. Less than half this group had sinus rhythm, and 1/4 had atrial fibrillation/flutter.
| Variable | Non-admitted deaths (n=63) | Admitted 30-day deaths (n=179) | Admitted 31-day to 1-year deaths (n=98) | p value | Admitted survivors at 1 year (n=300) |
|---|---|---|---|---|---|
| Left bundle branch block | 24 (38%) | 64 (36%) | 47 (48%) | 0.13 | 20 (7%) |
| Right bundle branch block | 11 (18%) | 19 (11%) | 2 (2%) | 0.004 | 9 (3%) |
| Anterior ST-elevation myocardial infarction † | 24 (62%) | 57 (50%) | 19 (37%) | 0.07 | 104 (37%) |
| Q-waves † | 13 (33%) | 57 (50%) | 28 (55%) | 0.11 | 93 (34%) |
| Conduction delay † | 21 (54%) | 30 (26%) | 8 (16%) | 0.0002 | 24 (9%) |
| Median number of leads with ST-elevation † (interquartile range) | 5 (3-7) | 4 (3-6) | 3 (2-5) | 0.004 | 4 (3-5) |
| Median maximum ST elevation † (mm) (interquartile range) | 4.5 (3-6) | 4 (2-5) | 3 (2-4) | 0.0008 | 3 (2-4) |
| Rhythm: | 0.0001 | ||||
| Sinus | 27 (43%) | 123 (69%) | 75 (77%) | 272 (91%) | |
| Atrial fibrillation/flutter | 16 (25%) | 23 (13%) | 12 (12%) | 14 (5%) | |
| Ventricular tachycardia | 4 (6%) | 6 (3%) | 4 (4%) | 0 (0%) | |
| Slow wide QRS complex ‡ | 7 (11%) | 3 (2%) | 0 (0%) | 2 (1%) | |
| Other | 9 (14%) | 24 (13%) | 7 (7%) | 12 (4%) |
† These analyses exclude left bundle branch block.
In striking contrast, in terms of co-morbidity characteristics associated with previous hospitalizations such as cardiovascular events and renal disease ( Table 3 ), group A patients had a profile resembling that of survivors. Compared with patients who died later, group A patients had a significantly lower prevalence of myocardial infarction, heart failure, and peripheral and cerebrovascular disease; they were also less likely to have had coronary revascularization. As well, they were less likely to have diabetes, arterial hypertension, and renal disease. Accordingly, the Charlson Comorbidity Index was the lowest and identical in group A patients and in survivors; it was significantly higher in those who died within 30 days and highest in patients who died in the 31-day to 1-year period.
| Variable | Group A | Group B | Group C | p value ∗ | Group D |
|---|---|---|---|---|---|
| Non-admitted deaths (n=63) | Admitted 30-day deaths (n=179) | Admitted 31-day to 1-year deaths (n=98) | Admitted survivors at 1 year (n=300) | ||
| Previous myocardial infarction | 7 (11%) | 30 (17%) | 23 (24%) | 0.12 | 33 (11%) |
| Previous percutaneous coronary intervention or coronary arterial bypass graft surgery | 2 (3%) | 12 (7%) | 10 (10%) | 0.23 | 20 (7%) |
| Previous congestive heart failure | 7 (11%) | 25 (14%) | 28 (29%) | 0.003 | 13 (4%) |
| Peripheral vascular disease | 6 (10%) | 35 (20%) | 20 (20%) | 0.15 | 22 (7%) |
| Cerebrovascular disease | 5 (8%) | 37 (21%) | 11 (11%) | 0.02 | 8 (3%) |
| Diabetes | 17 (27%) | 48 (27%) | 37 (38%) | 0.14 | 47 (16%) |
| Hypertension | 22 (35%) | 96 (54%) | 69 (70%) | <0.0001 | 155 (52%) |
| Renal disease | 6 (10%) | 55 (31%) | 35 (36%) | <0.001 | 22 (7%) |
| Median Charlson index (interquartile range) | 0 (0-2) | 2 (0-5) | 3 (1-6) | <0.0001 | 0 (0-2) |
| Any previous hospitalization | 39 (62%) | 112 (63%) | 73 (74%) | 0.10 | 118 (39%) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
