In a systematic province-wide evaluation of care and outcomes of ST elevation myocardial infarction (STEMI), we sought to examine whether a previously documented association between ambulance use and outcome remains after control for clinical risk factors. All 82 acute care hospitals in Quebec (Canada) that treated at least 30 acute myocardial infarctions annually participated in a 6-month evaluation in 2008 to 2009. Medical record librarians abstracted hospital chart data for consecutive patients with a discharge diagnosis of myocardial infarction who presented with characteristic symptoms and met a priori study criteria for STEMI. Linkage to administrative databases provided outcome data (to 1 year) and co-morbidities. Of 1,956 patients, 1,222 (62.5%) arrived by ambulance. Compared with nonusers of an ambulance, users were older, more often women, and more likely to have co-morbidities, low systolic pressure, abnormal heart rate, and a higher Thrombolysis In Myocardial Infarction risk index at presentation. Ambulance users were less likely to receive fibrinolysis or to be sent for primary angioplasty (78.5% vs 83.2% for nonusers, p = 0.01), although if they did, treatment delays were shorter (p <0.001). The 1-year mortality rate was 18.7% versus 7.1% for nonusers (p <0.001). Greater mortality persisted after adjusting for presenting risk factors, co-morbidities, reperfusion treatment, and symptom duration (hazard ratio 1.56, 95% confidence interval 1.30 to 1.87). In conclusion, ambulance users with STEMI were older and sicker than nonusers. Mortality of users was substantially greater after adjustment for clinical risk factors, although they received faster reperfusion treatment overall.
Highlights
- •
We compared ambulance users and nonusers with ST elevation myocardial infarction.
- •
Study patients arise from a complete system of care and 82 hospitals.
- •
Mortality was greater for users despite faster treatment and adjusting for risk.
- •
We extend this finding beyond single centers or voluntary registries and to 1 year.
- •
Awareness of how the patient arrived confers a simple means of risk stratification.
In patients with ST elevation myocardial infarction (STEMI), timely treatment with reperfusion, whether with primary percutaneous coronary intervention (PPCI) or fibrinolysis, is associated with better outcomes. Ambulance transport to hospital results in more timely treatment for patients with STEMI, but mortality of patients arriving by ambulance has been found to be greater than when they arrive by other means. Previous studies were not conducted within a complete system of care in a populous and large geographical region. In addition, they had limited sample size, short follow-up, examined a broader group of patients than those with STEMI, did not characterize outcomes, or only evaluated cardiac morbidity outcomes in the short term. Using data from a government-mandated province-wide evaluation of STEMI care, our objectives were to re-examine the association between ground ambulance use and patient outcomes up to 1 year after adjustment for clinical risk factors.
Methods
All 82 acute care hospitals that treated at least 30 acute myocardial infarctions annually in the province of Quebec, Canada (population 7.8 million persons in 2008), were included in our field evaluation of care, timing of treatment, and outcomes. This volume criterion allowed capture of >95% of all patients with myocardial infarction in Quebec. Our 6-month evaluation spanned the period October 1, 2008, to March 31, 2009. Data were abstracted retrospectively for the study cohort at each hospital by experienced medical record librarians on our research team, in collaboration with designated on-site librarians who received specific training. Data from the medical chart (and ambulance sheet, if applicable) were entered into a central secure Web site with automatic checks for missing and implausible values and independent validation of date-time values, as described previously. For patients transferred from a non-PPCI hospital to any of the 13 facilities capable of performing this procedure, data on PPCI were obtained from the second hospital. Patient identifiers were used to link medical chart data to hospital discharge and vital statistics databases to provide co-morbidities and outcomes. Ethics approval was granted by the Quebec Access-to-Information Commission.
Because the administrative code for STEMI used in the province during the study period had not been validated, a previously validated algorithm was used to identify the cohort. All patients had a hospital discharge date from October 1, 2008, to March 31, 2009; had a unique health insurance number (universal single-payer system); a final discharge diagnosis of myocardial infarction using the International Classification of Diseases, Tenth Revision (codes I21-I22.9 and I24); and presented at the first medical contact (to paramedic or at emergency department) with acute symptoms characteristic of myocardial ischemia. Of these patients, those considered as STEMI and thus included in this study had to fulfill one of the following criteria:
- 1.
They received fibrinolytic treatment within 4 hours of triage in the emergency department.
- 2.
They were sent for PPCI within 4 hours of emergency department triage and STEMI or left bundle branch block (LBBB) was mentioned in their medical chart. If STEMI or LBBB was not mentioned in the charts of patients sent for PPCI, they were included if STEMI or LBBB was confirmed by at least 1 of 2 cardiologists on the research team using the first available electrocardiogram from the receiving emergency department and any prehospital electrocardiograms. We chose a 4-hour time window for (1) and (2) because pilot tests of our field evaluation methods showed it was highly unlikely for patients with STEMI to be treated with fibrinolysis or to be sent for PPCI beyond this period.
- 3.
They did not fulfill (1) or (2) but STEMI or LBBB was documented in their medical chart and was confirmed by electrocardiographic review by at least 1 of the study cardiologists.
The principal outcome measure was death from any cause at 1 year after emergency department triage. We also report death in hospital and at 30 days, readmission to hospital for myocardial infarction or for heart failure at 1 year, and a composite 1-year outcome consisting of death, readmission for myocardial infarction, or readmission for heart failure. The following time variables were measured, according to American College of Cardiology/American Heart Association standard definitions: (1) time from onset of symptoms to emergency department triage, (2) time from emergency department triage to first electrocardiogram (door to ECG), (3) time from triage to administration of fibrinolysis (door to needle), and (4) time from triage to first use of a device with intention to perform PPCI (door to balloon).
Clinical patient factors included age, gender, symptom duration (before triage), triage during regular working hours (or not), earliest measured systolic blood pressure (at first medical contact), and heart rate at first electrocardiogram (categories for the last 2 variables are listed in Table 1 ). The Thrombolysis In Myocardial Infarction (TIMI) risk index, a previously validated predictor of mortality in patients with STEMI, was calculated according to the formula: (heart rate × [age/10] 2 )/systolic blood pressure. Anterior wall STEMI was identified by the study cardiologists on the basis of the presenting electrocardiogram. Using the medicoadministrative hospitalization database, presence of peripheral vascular disease, hypertension, renal disease, chronic obstructive pulmonary disease, cancer, and diabetes was assessed at the time of the index myocardial infarction admission or if specified as a discharge diagnosis of any hospital stay in the previous 5 years. To avoid misclassifying a complication of the index myocardial infarction event as a co-morbidity, heart failure and arrhythmia were only included as co-morbidities if they were noted as diagnoses in preceding (5 years) hospitalizations. For previous myocardial infarction, a discharge diagnosis had to be noted in the period from 5 years to 28 days before the index event.
| Characteristic | Ambulance Users (N = 1222) | Non-Users (N = 734) | p |
|---|---|---|---|
| Median age (25 th –75 th percentile; years) | 65 (54–78) | 60 (51–68) | <0.001 |
| Women | 398 (32.6%) | 152 (20.7%) | <0.001 |
| Anterior wall STEMI | 300 (24.5%) | 177 (24.1%) | 0.83 |
| Heart rate >100 or <60 beats/min | 442 (36.2%) | 200 (27.2%) | <0.001 |
| Systolic blood pressure <90 mm Hg | 118 (9.7%) | 17 (2.3%) | <0.001 |
| Median TIMI risk index (25 th –75 th percentile) | 24.6 (16.4–38.0) | 18.1 (13.4–25.8) | <0.001 |
| Symptoms to triage ≤3 hours | 755 (61.8%) | 408 (55.6%) | 0.01 |
| Hospital arrival within 8 am–6 pm | 652 (53.4%) | 412 (56.1%) | 0.23 |
| Myocardial infarction | 205 (16.8%) | 79 (10.8%) | <0.001 |
| Heart failure ∗ | 87 (7.1%) | 28 (3.8%) | 0.003 |
| Peripheral vascular disease | 125 (10.2%) | 42 (5.7%) | 0.001 |
| Arrhythmia | 87 (7.1%) | 32 (4.4%) | 0.01 |
| Hypertension | 607 (49.7%) | 324 (44.1%) | 0.02 |
| Cerebrovascular disease | 74 (6.1%) | 33 (4.5%) | 0.14 |
| Renal disease | 172 (14.1%) | 62 (8.4%) | <0.001 |
| Chronic obstructive pulmonary disease | 207 (16.9%) | 96 (13.1%) | 0.02 |
| Cancer | 80 (6.6%) | 44 (6.0%) | 0.63 |
| Diabetes mellitus | 241 (19.7%) | 129 (17.6%) | 0.24 |
∗ Includes heart failure, pulmonary edema, or cardiogenic shock.
Continuous data are reported as medians with twenty-fifth and seventy-fifth percentiles. Differences in proportions were tested with the 2-sided Pearson’s chi-square test, whereas differences in medians were tested with the 2-sided Wilcoxon rank sum test. A Cox proportional hazards model was used to determine the independent association of ambulance use with 1-year mortality after adjustment for clinical risk factors (covariates). We a priori selected the covariates based on previous studies in literature. We used a logarithmic transformation of the TIMI risk index to preserve its linearity with mortality. Because ambulance use varied by region, we used a generalized estimating equation, designating 16 clusters for the health administrative regions, to estimate the parameters.
In a second analysis, we added 2 time-varying interaction terms to the model to explore whether the hazards of death associated with (1) ambulance use and (2) reperfusion treatment (fibrinolysis or sent for PPCI) were greater in the acute follow-up period (≤30 days) compared with the later period (up to 1 year). Finally, we selected only those patients who actually received reperfusion therapy (fibrinolysis or PPCI) and ran the first mortality model (i.e., without interaction terms) with 2 additional variables: timeliness of treatment (door to needle ≤30 minutes or door to balloon ≤90 minutes ) and treatment approach (fibrinolysis, direct admission to a PPCI center, or transfer from a non-PPCI center for PPCI, with the last of these coded as the reference group). Data were analyzed using SAS software (SAS Institute Inc, Cary, North Carolina).
Results
Of 8,047 patients with a discharge diagnosis of myocardial infarction in the 82 hospitals, we identified 1,956 patients with STEMI in the 6-month period of the study; of the latter, 1,222 patients (62.5%) arrived by ambulance. Charts were available and linkage to administrative databases was possible for all study patients. Table 1 compares characteristics of patients who did and did not use an ambulance. Compared with patients not arriving by ambulance, ambulance users were significantly older, more likely to be women, and more likely to have previous myocardial infarction, heart failure, peripheral vascular disease, arrhythmia, hypertension, renal failure, and chronic obstructive pulmonary disease. Ambulance users were more likely to have low systolic pressure, abnormal heart rate, and a higher TIMI risk index than nonusers. A greater proportion of ambulance users were triaged at the emergency department within 3 hours of symptom onset. However, similar proportions of patients in the 2 groups had an anterior wall STEMI, were triaged during regular working hours, and had a history of cerebrovascular disease, cancer, or diabetes. Table 2 compares treatment characteristics for ambulance users and nonusers. The proportion of patients who received fibrinolysis or were sent for PPCI within 4 hours was lower in the ambulance group. However, ambulance users had shorter delays to first in-hospital electrocardiogram and were more likely to have timely reperfusion therapy.
| Variable | Ambulance Users (N = 1222) | Non-Users (N = 734) | p |
|---|---|---|---|
| Received fibrinolysis | 154 (12.6%) | 141 (19.2%) | <0.001 |
| Sent for PPCI at first hospital (direct admission PPCI) | 361 (29.5%) | 129 (17.6%) | |
| Transferred for PPCI from non-PPCI hospital (transfer PPCI) | 444 (36.3%) | 341 (46.5%) | |
| Any of the above (fibrinolysis or referral for PPCI) | 959 (78.5%) | 611 (83.2%) | 0.01 |
| Door-to-ECG ≤10 min | 741/1200 (61.8%) | 333/731 (45.6%) | <0.001 |
| Median time, min (25 th –75 th percentile) | 8 (2–15) | 11 (5–19) | <0.001 |
| Door-to-needle ≤30 min | 91/154 (59.1%) | 51/141 (36.2%) | <0.001 |
| Median time, min (25 th –75 th percentile) | 25 (19–45) | 39 (26–55) | <0.001 |
| Door-to-balloon ≤90 min (direct admission PPCI) ∗ | 249/340 (73.2%) | 72/121 (59.5%) | 0.005 |
| Median time, min (25 th –75 th percentile) | 66.5 (48–93) | 87 (63–103) | <0.001 |
| Door-to-balloon ≤90 min (transfer PPCI) ∗ | 109/382 (28.5%) | 67/301 (22.3%) | 0.06 |
| Median time, min (25 th –75 th percentile) | 104 (87–134) | 116 (94–147) | <0.001 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
