All patients with a definitive diagnosis of STEMI included in the AMIS Plus registry from January 1997 to December 2017 were screened. For the present study, STEMI was defined by characteristic symptoms of acute myocardial ischemia, persistent ST-segment elevation on the initial electrocardiogram, and elevation of cardiac biomarkers. ST-segment elevation was defined as follows: at least 2 contiguous leads with ST-segment elevation of 2.5mm in men aged <40 years, 2mm in men aged ≥ 40 years, or 1.5mm in women in leads V2–V3 and/or 1mm in the other leads. Concerning biomarkers, diagnosis of STEMI required an increase in creatinine kinase-MB serum level at least twice above the upper limit of normal and/or an increase in troponin (I or T) serum level above the upper limit of normal; for each biomarker, we used individual hospital cutoffs for diagnosis of myocardial infarction. For each included patient, diagnosis conformed to the prevailing guidelines in use at the time of inclusion.

Patient-related delay (ie, time from symptom onset to first medical contact leading to hospitalization) was then assessed and STEMI patients were classified as early-comers (ie, patients with a symptom-to-door time ≤ 12hours) or latecomers (ie, patients with a symptom-to-door time> 12hours). Of note, both “timing of symptom onset” and “timing of first medical contact leading to hospitalization” were systematically included in the AMIS Plus questionnaire: patient-related delay was determined a posteriori based on these data. The subsequent analyses were essentially focused on latecomer STEMI patients. For this patient subgroup, baseline clinical features were assessed, including all classic cardiovascular risk factors, previous coronary history, and clinical status at admission (Killip class, blood pressure and heart rate at hospitalization, and resuscitation prior to hospital admission). To account for comorbidities, we calculated the Charlson comorbidity index.16 We collected data on immediate drug therapy (defined as therapy administered within the first 24hours of hospitalization), revascularization therapy during the index hospitalization (including door-to-balloon time in patients presenting between 12 and 48hours after symptom onset and undergoing PCI, which was defined as the time interval between first medical contact leading to hospitalization and crossing of the culprit lesion with a guidewire), and drug therapy at discharge to home.

All treatments were prescribed according to current practice and guideline recommendations at the time of the index presentation and treatment choice was left to the discretion of treating physicians. Concerning procedural data, only infarct location and the number of diseased vessels were available in a sufficient number of patients and were, therefore, included in the present analysis. Concerning outcomes, data on mortality, cardiogenic shock, reinfarction, significant bleeding events, and cerebrovascular events during the index hospitalization were collected. The methodology to assess outcomes was uniform during the study. In line with the Killip definition, cardiogenic shock was defined as hypotension with symptoms and/or signs of hypoperfusion. Reinfarction was defined as clinical signs or symptoms of ischemia with ECG changes indicative of new ischemia and a new rise in cardiac biomarkers following the initial infarction. Bleeding complications were recorded if deemed clinically relevant by the individual physician in charge of the patient. Stroke was defined as any event due to ischemic, thrombotic or hemorrhagic disturbance confirmed by a neurologist or an imaging modality.

The study endpoint was analysis of the temporal trend in in-hospital mortality in latecomer STEMI patients. In these patients, we also evaluated temporal trends in prevalence, in-hospital and discharge treatments, and other major in-hospital outcomes. Moreover, we aimed to identify independent predictors of in-hospital mortality in latecomer STEMI patients presenting between 12 and 48hours from symptom onset (hereafter identified as 12- to 48-hour latecomers).

Categorical variables are presented as percentages. Continuous, normally distributed variables are expressed as the mean±1 standard deviation, while continuous, nonnormally distributed variables are expressed as the median [interquartile range]. Temporal trends were tested using the linear-by-linear test. To assess independent predictors of in-hospital mortality, multivariate analysis was performed with a logistic regression model. All variables included in the final logistic regression model were selected based on known clinical significance or on the results of preliminary age-adjusted logistic regression models. Globally, variables to be entered in the final model were carefully chosen, based on the number of mortality events, to ensure parsimony of the statistical model. To account for the potential confounding effect of changes over time (including variations in pharmacological treatments), we entered the variable “admission period” (as a 3-year period) in the final multivariate analysis. Model fit was assessed using the Hosmer-Lemeshow test. Odds ratios (OR) are reported with 95% confidence intervals (95%CI). A P value of less than .01 was considered significant. IBM SPSS Statistics (version 23, IBM Corp. Armonk, New York, United States) was used for statistical analyses.

Overall, 27 653 STEMI patients were identified. A total of 422 patients were excluded because of missing data on patient-related delay and thus 27 231 patients were available for the final analysis. Of these, 22 928 patients were early-comers (84%) and 4303 were latecomers (16%). During the study period, the prevalence of late presentation significantly decreased from 22% to 12.3% (P <.001; figure 1).

Figure 1.

Prevalence of late presentation during the 20-year study period. Data reported as percentages. STEMI, ST-segment elevation myocardial infarction.

(0.11MB).

Patient-related delay did not significantly decrease over time within the overall latecomer population (data not shown, P=nonsignificant). However, when we focused only on 12- to 48-hour latecomers (3704 patients, 85% of the overall latecomer population), patient-related delay significantly decreased from 1260 [962-1635] minutes to 1103 [880-1897] minutes (P <.001; figure 2). Similarly, patient-related delay significantly decreased over time in the overall STEMI population from 210 [114-510] minutes to 158 [90-352] minutes (P <.001).

Figure 2.

Trend in patient-related delay (in minutes) during the 20-year study period.

(0.22MB).

During the 20-year study period, there was a significant decrease in the prevalence of female sex from 30.5% to 24.2% (P=.005). The proportion of patients with hypertension and dyslipidemia significantly increased. Conversely, we observed a significant decrease in the proportion of patients with diabetes and a previous history of coronary artery disease or acute myocardial infarction (AMI). There was also a significant reduction in the proportion of patients with resuscitation prior to admission or presenting in advanced Killip class (> 2). Trends in baseline clinical features are summarized in table 1.

Concerning immediate pharmacological treatments, there was a significant increase in the use of P2Y12 inhibitors (17.7%-91.5%, P <.001), statins, and renin-angiotensin-aldosterone system blockers during the study period. Conversely, in-hospital use of beta-blockers and glycoprotein IIb-IIIa inhibitors significantly decreased. In 12- to 48-hour latecomers, the PCI rate markedly increased from 11.9% to 87.9% (P <.001). In this patient subgroup, door-to-balloon time significantly decreased from 110 [31-327] minutes to 58 [27-116] minutes (P <.001). Referral for coronary artery bypass grafting in the whole latecomer cohort did not significantly change over time. Data on temporal trends in in-hospital treatments are shown in table 2.

At discharge to home, there was a significant increase in the prescription of aspirin (80.9%-96%), P2Y12 inhibitors (6%-90.7%), statins (55.1%-93.5%), and renin-angiotensin-aldosterone system blockers (60.8%-84.2%) (P <.001 for all variables). Trends in discharge treatments are summarized in table 3.

During the 20-year study period, in-hospital mortality decreased markedly from 12.4% to 4.5% (P <.001). The prevalence of both cardiogenic shock (14.6%-4.3%) and reinfarction (5.4%-0.2%) developing during the index hospitalization significantly decreased (P <.001 for both variables). Conversely, the prevalence of cerebrovascular events remained fairly stable and there was a strong trend toward a significant increase in the prevalence of bleeding complications (2.1%-5.1%, P=.025). Length of hospitalization in acute care facilities significantly decreased from 10 [6-14] days to 4 [1-7] days (P <.001). Trends in major in-hospital outcomes are shown in detail in table 4.

Anterior infarct location and multivessel disease had no impact on in-hospital mortality as assessed in preliminary age-adjusted models and were, therefore, not included in the final logistic regression model (OR, 0.95; 95%CI, 0.65-1.41 and OR, 1.59; 95%IC, 0.90-2.79, respectively; P=nonsignificant for both variables). Similarly, heart rate at hospitalization showed no significant correlation with in-hospital mortality in a preliminary age-adjusted model that also included systolic blood pressure and Killip class at hospitalization and was, therefore, excluded from the final analysis (OR, 1.00; 95%CI, 0.99-1.01; P=.772). In our final logistic regression model, there was a slight but significant direct correlation between age and in-hospital mortality (OR, 1.07; 95%CI, 1.07-1.11; P <.001). Advanced Killip class (> 2) at hospitalization was the strongest independent predictor of in-hospital mortality (OR, 5.91; 95%CI, 2.79-12.53; P <.001). PCI had an impressive independent protective effect on in-hospital mortality with an OR of 0.29 (95%CI, 0.15-0.55; P <.001). Of note, PCI had no protective effect on mortality in patients presenting more than 48hours after symptom onset (OR, 0.41; 95%CI, 0.08-2.04; P=.123). Data from the multivariate analysis are summarized in table 5.

Of interest, neither a door-to-balloon time below 60minutes nor a door-to-balloon time below 90minutes had a significant impact on in-hospital mortality as assessed in 2 distinct age-adjusted models (OR, 0.71; 95%CI, 0.40-1.26; P=.247 and OR, 0.72; 95%CI, 0.42-1.24; P=.236, respectively); cutoffs for the analysis were chosen based on the latest recommendations on STEMI by the European Society of Cardiology.12 Moreover, when we assessed the impact of PCI on in-hospital mortality in an age-adjusted model including total ischemic time (defined as the sum of patient-related delay and door-to-balloon time), PCI was still protective, even if it showed only a strong trend in terms of statistical significance (OR, 0.36; 95%CI, 0.14-0.93; P=.035).

Our study has some limitations. First, the patients were treated at many different hospitals, including centers without PCI facilities, and the choice of interventions and medications was left to the discretion of the treating physicians, which could have caused an unknown bias. However, our data, drawn from a real-life scenario, help to provide a better understanding of the overall burden of late presentation. Moreover, in our study, patient-related delay was defined as “time from symptom onset to first medical contact leading to hospitalization”; this definition slightly differs from the that used in latest guidelines of the European Society of Cardiology on STEMI, which define patient-related delay as “time from symptom onset to first medical contact”.12 We acknowledge that this slightly different definition could have led to an overestimation of delay in the patient subgroup that first alerted the prehospital network. However, in Switzerland, there is an efficient prehospital emergency network, whose activation times are usually very short (fibrinolysis progressively disappeared from national algorithms for the treatment of STEMI). Moreover, the median time of presentation in our latecomer population was well above 12hours throughout the study period. Therefore, we believe that this slight difference in the definition of patient-related delay did not have a major impact on our results. Furthermore, it is not known whether patients were still experiencing symptoms and/or signs suggestive of ongoing ischemia at hospital admission. However, previously published trials showed that patients with ongoing signs of ischemia at hospital admission usually represent only a minority of latecomer STEMI patients, around 15%.8 Therefore, the observed improvements in prognosis more likely reflect a wider clinical benefit of PCI. In our study, we could not account for all clinical, anatomic, and procedural variables potentially influencing clinical outcomes. However, this analysis included most important variables affecting short-term prognosis in STEMI patients. Finally, the lack of data on long-term outcome represents an additional limitation of our study. In AMIS Plus registry follow-up data, obtained through a telephonic interview, are available only for a minority of patients and did not therefore allow meaningful analysis of temporal trends in 1-year mortality or of independent predictors of long-term outcome in our latecomer population.