Timely reperfusion therapy is the most important component of the treatment for ST-segment elevation acute myocardial infarction (STEMI), as it strongly influences short- and long-term clinical outcome.1 2 The aim of this strategy is to achieve early and complete recanalisation of the infarct-related artery, in order to stop or reduce the progression of ischaemic myocardial damage.

As compared with fibrinolytic therapy (FT), primary percutaneous coronary intervention (pPCI) has been associated with reduction of death, reinfarction, intracranial bleeding, reocclusion of the infarct artery and recurrent ischaemia.3 Therefore, pPCI should be considered the preferred reperfusion strategy when applicable without excessive delay, as opposed to FT. It should be performed by an expert interventional cardiologist, and when the patient shows well-defined, high-risk clinical characteristics or absolute contraindications to FT.1 2 In daily practice, however, pPCI cannot be offered to all patients presenting with acute STEMI, primarily because of logistical constraints or long distances from the first available catheterisation laboratory. Hence, the choice of reperfusion treatment must take into account a number of patient and logistical variables.4 In addition, despite the overwhelming evidence of clinical efficacy, many studies continue to indicate that reperfusion therapy as a whole is underused and often not administered within an acceptable time delay after presentation.5

To increase the rates of reperfusion therapy, provide the best available treatment to all patients, reduce the delay in time to treatment and, ultimately, improve the clinical outcome, current guidelines recommend the institution of a territorial organisation which comprises prehospital triage and development of an interhospital network for patients with STEMI based on a hub-and-spoke model.1 2

The aim of this study was to assess the clinical impact of the development of a cooperating network for STEMI between the emergency medical service (EMS), coronary care units (CCUs) and cardiac catheterisation laboratories (CCLs) within a large provincial area.

METHODS

Organisation of the territorial network

In 2003, immediately after the publication of new guidelines for management of patients presenting with STEMI by the European Society of Cardiology,1 the Health Care Agency of the Italian region Emilia-Romagna (four million inhabitants) promoted the development of a territorial network for the treatment of STEMI. The regional project had to be gradually implemented at a provincial level (nine provinces) through the definition of new diagnostic and therapeutic pathways. The realisation of the project was delegated to provincial committees of experts, composed of cardiologists and 118 EMS operators, and was led by the Regional Cardiology Committee, which was a task force composed of cardiologists and health administrators.

In the province of Bologna (around one million inhabitants) there are 13 hospitals admitting patients with STEMI (fig 1), with one high-volume catheterisation laboratory available for 24 h 7 days a week, one catheterisation laboratory with 12-h pPCI facilities (only daytime, from 8 am to 8 pm, including weekends), 5 CCUs, and 7 peripheral hospitals without CCU. Another catheterisation laboratory with 24 h 7 days a week availability in an adjacent province was available as a hub for one of the peripheral hospitals (Imola). The emergency calls from the province area were received by the 118 dispatch centres of Bologna, which coordinates all the resources of the EMS. Telemedicine with ECG transmission from the ambulance through a GSM mobile phone, not available in 2002, was implemented in the vast majority of the territory.

The provincial committee considered all the resources available and, accordingly, recommended pPCI to the following patients: (a) patients with STEMI of <12 h presenting at the emergency department of a PCI centre; (b) patients triaged by the 118 ambulance system within 12 h from symptom onset and within 60 min driving time from a PCI centre; (c) high-risk STEMI in patients who spontaneously presented at local hospitals; (d) patients with absolute contraindications to FT. Lytic therapy, preferably started in the ambulance, and immediate transfer to a CCU was recommended for all other patients with STEMI presenting within 12 h from symptom onset. Transfer for rescue PCI was encouraged for patients with evidence of persistent ST elevation and pain after the administration of FT. Rapid transfer back to the CCU of non-PCI hospitals was recommended for patients shortly after pPCI, generally after a few hours in the CCUs of hub centres and preferably within 24 h, provided that no serious complications had arisen.

Patient population

We compared the clinical outcome of all patients with STEMI admitted to any hospital of the province of Bologna during the year 2004 (first year of good network implementation), regardless of the time of presentation, with the outcome of all patients with STEMI admitted in 2002, the last year before the network started. Hospital discharge records were reviewed to identify patients admitted with a diagnosis of myocardial infarction according to the ICD-9-CM classification (code 410.X). All clinical files were subsequently checked by two independent clinicians in order to identify accurately those with a definite diagnosis of STEMI and to collect additional data about prehospital resources use. All patients undergoing pPCI were also prospectively enrolled in a dedicated database, which was investigated for additional information.

Definitions and follow-up

The diagnosis of STEMI at admission was based on the presence of chest pain lasting for >20 min, and the presence of significant ST-segment elevation (⩾0.1 mV in two adjacent leads if leads I–III, aVF, aVL, V4–V6, and ⩾0.2 mV in leads V1–V3), as recorded in the first ECG obtained. According to the Italian STEMI Consensus document,6 high risk was defined on the basis of at least one of the following criteria: heart failure (Killip class >1), systolic blood pressure <100 mm Hg and heart rate >100/bpm, age >75 years, ST elevation in ⩾6 ECG leads. The Charlson comorbidity index was calculated with the Romano adaptation.7 8 Cardiogenic shock was defined as persistent systolic blood pressure <90 mm Hg, or need for inotropes or intra-aortic balloon pump to maintain a pressure >90 mm Hg. Chronic renal failure was defined by the presence of dialysis or previous hospitalisation owing to chronic renal failure. The primary evaluation of the survey was in-hospital mortality. As secondary objectives, we assessed the incidence of major adverse cardiac and cerebrovascular events (MACCE), defined as (a) death (cardiac and non-cardiac); (b) non-fatal acute myocardial infarction (MI); (c) stroke and (d) coronary revascularisation procedures (both coronary artery bypass graft and PCI over 1-year follow-up. MI during follow-up was diagnosed by local cardiologists at the hospital of admission. We also evaluated the rate of rehospitalisation for coronary heart disease (including angina, MI, arrhythmias, heart failure) and implantation of a pacemaker or automatic implantable cardioverter defibrillator.

Follow-up was obtained directly and independently from the Regional Health Agency through the analysis of the hospital discharge records and the mortality registries. This ensures a complete follow-up for 100% of patients resident in the region, including all out-of-hospital deaths, and all rehospitalisations, even when these occurred in different regions. All repeat interventions during follow-up were prospectively collected from the individual institutions as well. Hospital records were reviewed for additional information whenever deemed necessary.

Statistical analysis

Continuous variables were expressed as mean (SD) and were compared using the Student unpaired t test. Categorical variables were expressed as counts and percentages, and a χ² test or Fisher’s exact text were used for comparison. The cumulative incidence of adverse events was estimated according to the Kaplan–Meier method and compared using the log-rank test. A multivariable analysis was performed to assess the odds ratios for adverse events associated with the following variables: year of treatment, age, gender, diabetes mellitus, previous MI, previous PCI, previous coronary artery bypass graft, heart failure, peripheral vascular disease, cerebrovascular disease, renal failure, malignancy, shock, anterior location. The logistic model showed good predictive value (C statistic = 0.848), and good calibration characteristics using the Hosmer–Lemeshow test (p = 0.58). All analyses were performed with the SAS 8.2 system.

RESULTS

During the study period, 1823 patients in the province were admitted with a definite diagnosis of STEMI (n = 858 in 2002, n = 965 in 2004), with a corresponding incidence per million residents equal to 940 in 2002 and 1032 in 2004 (an increase of 11%). Baseline patient characteristics (table 1) were not significantly different between the 2 years under investigation, with the exception of anterior location (44.3% in 2002 vs 49.2% in 2004, p = 0.04).

Figure 1 illustrates the change in reperfusion therapy strategy for patients admitted within 12 h from symptom onset in different areas of the provincial territory. The use of resources was remarkably different: patients admitted in 2004 reached the hospital more often through the 118 EMS (46.5% 2002 vs 59.3%, 2004; p<0.001). In 2004, 24.1% of the patients were admitted directly from the EMS to the CCL, and 15.3% from the EMS directly to the CCU, both options which were not available in 2002. In addition, patients had more frequently the ECG performed in the territory (n = 281 (32.8%) in 2002 vs n = 510 (52.8%) in 2004, p<0.001), and 62% of these ECGs were transmitted through the newly activated telemedicine service to a cardiologist on duty in a CCU. The percentage of patients transferred from spoke to hub hospitals for pPCI increased from 0.3% in 2002 to 6.8% in 2004 (p<0.001). Patients with at least one interhospital transfer increased from 22% in 2002 to 34% in 2004 (p = 0.005), attesting to enhanced collaboration between hospitals. No deaths occurred during interhospital transfer.

Figure 2 shows the use of reperfusion therapy. Between 2002 and 2004 there was a significant decrease in the use of FT (38.2% vs 10.7%, p<0.001), a striking increase in the use of pPCI (20.2% vs 65.6%, p<0.001), and a major decrease in the rate of patients not undergoing any reperfusion treatment (from 41.6% to 23.7%, p<0.001). When considering only patients admitted within 12 h, the reperfusion rate was 68.7% in 2002 and 89.8% in 2004 (p<0.001). Table 2 reports the baseline characteristics of the patients according to reperfusion strategy in the 2 years, and fig 3 shows the mortality rates. Among patients undergoing reperfusion treatment, FT accounted for 65.5% in 2002 and 14.1% in 2004, whereas pPCI was respectively 34.5% and 85.9%. Among patients treated with lytic agents, fibrin-specific lytic drugs were used in 92.6% of the patients admitted in 2002 and 96.2% in 2004 (p = 0.21). Rate of prehospital FT increased after the implementation of the network (4.0% in 2002 vs 23.3% in 2004; p<0.001), as well as use of rescue PCI (14.6% in 2002 vs 32.0% in 2004, p<0.001) after failed FT. Late reperfusion with PCI (ie, >24 h after symptom onset) was carried out in an additional 53 patients (6.2%) in 2002 and 63 patients (6.5%) in 2004.

Table 3 reports the characteristics of the patients receiving reperfusion therapy as compared with patients who did not. Not surprisingly, the clinical profile of the latter patients was clearly worse. The reasons for not receiving reperfusion were STEMI recognised late (atypical symptoms, minimal or absent ECG changes), contraindication to reperfusion (eg, associated cerebrovascular accident, continuing haemorrhage), extremely compromised clinical conditions (eg, terminal neoplasia or dementia).

In-hospital mortality decreased from 17.0% in 2002 to 12.3% in 2004 (p = 0.005). Median hospital stay was 9 days in 2002 (interquartile range 6–12) and 7 days in 2004 (interquartile range 5–10; p<0.001). Table 4 lists the factors significantly associated with in-hospital death.

One-year follow-up was completed for 100% of the patients resident in the region (n = 1714, 94% of all patients). The incidence of mortality was 23.9% for the 2002 group and 18.8% for the 2004 group (p = 0.009) (fig 4). All MACCE were significantly less common in the year 2004 cohort (39.5% vs 34.3%, p = 0.01) (table 5).

DISCUSSION

The main finding of this study is that implementation of a regional reperfusion strategy effectively decreased rates of patients who were “eligible but untreated” and was associated with reduced in-hospital and mid-term mortality.

The level of this reduction goes well beyond the expected advantage of pPCI over FT,3 reflecting several improvements in the diagnostic and therapeutic processes: (a) increased rates of prehospital diagnosis and triage; (b) implementation of telemedicine for ECG transmission; (c) a major shift from FT to pPCI as the preferred reperfusion strategy; (d) development of a fast-track route from emergency ambulance to catheterisation laboratory; (e) implementation of prehospital FT, especially with long transportation times to the hub, and centralisation of patients; (f) interhospital transfer through the 118 EMS; (g) increase of rescue PCI rate; (h) overall increase in patients receiving reperfusive therapy.

Although pPCI is considered better than FT, data from the National Registry of Myocardial Infarction indicate that fewer than half of patients are treated in time frames recommended by international guidelines,5 and this figure is worse for patients undergoing pPCI after interhospital transfer.9 10 A Bayesian meta-analysis of 22 published trials confirmed that the clinical benefit of pPCI over FT is inversely related to the former’s additional time delay,11 and pPCI would be cost effective, according to the British threshold, only for delays of up to about 80 min.12 Hence, two key steps to optimise the treatment of STEMI are to improve the accessibility of pPCI within the healthcare system, and to keep the door-to-balloon time within an acceptable time delay, which are the principal objectives of territorial networks for STEMI.13^–16 Accordingly, use of pPCI in our province increased from 20.2% to 65.6%, and time to reperfusion was reduced by the implementation of prehospital diagnosis and triage. We previously reported the impact of the newly implemented ambulance-based prehospital diagnosis of STEMI followed by direct referral to the CCL.17 This strategy was associated with >45 min reduction in treatment delay, and a better clinical outcome especially in high-risk patients such as those with cardiogenic shock.17 Similarly, more recent studies showed that door-to-balloon-times <90 min were more often achieved when trained paramedics independently triaged and transported patients directly to a designated pPCI centre than when patients were referred from emergency departments.18 19 Telemedicine that uses mobile phone lines to send field ECGs to a cardiologist in the CCU for interpretation is an important option for better integration between the hospitals and the territory.17 20 21 Of note, we previously showed that when the pPCI procedure is performed at a centre specifically dedicated to STEMI treatment within a well-organised regional network, patients treated during “off-hours” had similar in-hospital and 1-year mortality rates to those of patients admitted during normal working hours.22 The results of our registry confirm and expand the results of the Vienna registry,15 where a network between the Viennese Ambulance Systems and five high-volume interventional cardiology departments was created. The authors reported a significant increase in patients receiving reperfusion (from 66% to 86.6%) and a significant decrease in in-hospital mortality (from 16% before establishment of the network to 9.5%).15

Rates of rescue PCI in our network increased remarkably. Timely performed rescue PCI has been associated with improved regional infarct-zone wall motion and greater freedom from adverse in-hospital events than a deferred PCI strategy or medical treatment.23 24 It is therefore recommended for patients who develop shock, heart failure, haemodynamic or electrical instability, or persistent ischaemic symptoms.1 2 Conversely, late reopening of the infarct-related artery in patients who are stable does not provide clinical benefit and might even be harmful.25 26 This highlights the importance of interhospital transfer in the acute phase of MI at least for high-risk patients for whom FT has failed.

Prehospital FT has been associated with reduced time to FT by almost 1 h and a significantly lower adjusted long-term mortality than with regular in-hospital FT.27 Thus, it should be considered as the preferred option when the reperfusion strategy is FT. After implementation of the STEMI network, we observed a noteworthy increase of prehospital FT in our province (from 4.0% to 23.3%), which certainly contributed to the improvement in clinical outcome. Although not statistically significant, prehospital FT was associated with lower in-hospital mortality (2.7%) than standard in-hospital FT (7.6%) (p = 0.17).

The rate of STEMI per million inhabitants in our registry was higher than expected, probably because of the inclusion of patients not admitted to a CCU and patients in the subacute phase of the event. Nevertheless, the increased incidence of STEMI (+11%) between the two study periods is mysterious, because it conflicts with the declining STEMI rates reported elsewhere. We found two possible explanations, both of which are speculative: (a) diffusion of information through the general practitioners, the mass media and the individual patient’s experience decreased the number of missed diagnoses; (b) an increased number of ambulances and trained personnel on the territory reduced the number of patients who died before admission to hospital.

Another positive effect of the network was the reduction of length of hospital stay, although it remained considerably longer than that seen by other authors (ie, median 3 days in the Minnesota network as compared with 7 days in our experience).16 Probably, the difference is attributable to inclusion of patients after 12–24 h from symptom onset and patients with important underlying comorbidities, which were excluded from other reports. Nevertheless, we think that hospital stay could be further reduced by promoting the early discharge of patients without complications after pPCI.

Finally, although rates of reperfusion could be further improved, as could many other aspects of the network, we showed that organisation of the territorial STEMI network improved the overall process of care, and allowed us to meet the main objective of achievement of at least 75% of reperfusion therapy according to the policy of the European Society of Cardiology.28 Importantly, key organisational details should be periodically examined, and monitoring of quality is mandatory. Different network models can be developed for different areas. Several financial, political and geographical problems will affect which model is more feasible for a particular location.

Limitations

This study suffers the inherent limitations of all observational studies. On the other hand, the unrestricted inclusion of all patients admitted with the same diagnosis, in the same territorial area and hospitals, in a virtually subsequent time frame, gives a very comprehensive view of the final effects of the network. It also overcomes the limitation of voluntary participation of centres in a registry, and limitations due to observations from a single perspective (eg, perspective limited to hub hospitals). We did not provide precise information about time to treatment, because of the lack of accurate information for FT-treated patients. However, we described a major increase of prehospital FT that could be considered as a surrogate for shorter time to treatment in this setting, and changes in time to reperfusion for hub hospitals in our network have been previously reported.17

CONCLUSION

The organisation of regional STEMI networks to facilitate access to rapid mechanical reperfusion at high-volume centres improves the overall quality and consistency of STEMI patient care, and reduces in-hospital and mid-term mortality. The key point of the network is the multidisciplinary cooperation between the EMS, emergency rooms, CCU and CCL cardiologists, with the coordination and support of healthcare government agencies.