In the PROGALIAM protocol, the northern region of the Galician STEMI network includes the areas of Ferrol, Lugo, and A Coruña. This region has a population of approximately 1 million inhabitants. The PPCI referral hospital for the region is the Complejo Hospitalario Universitario A Coruña. The region also contains 4 non-PPCI hospitals. Of these, 2 are level 2 centers, one in Ferrol (Hospital Arquitecto Marcide) and the other in Lugo (Hospital Univesitario Lucus Augusti). The other 2 are district hospitals, one in the extreme west of the region and the other to the north of Lugo. The organization of the STEMI network in northern Galicia is presented in figure 1, including distances between hospitals and the quantity and location of available specially equipped ambulances.

Figure 1.

The northern Galicia STEMI-treatment region, showing the the reference population, hospitals, specialized emergency services, and study periods examined in the IPHENAMIC study. CHUAC, Complejo Hospitalario Universitario de A Coruña; H, hospital; HAM, Hospital Arquitecto Marcide; HULA, Hospital Universitario Lucus Augusti; PPCI, primary percutaneous coronary intervention; STEMI, ST-segment elevation myocardial infarction.

(0.33MB).

The participating hospitals were asked to provide a list of all events registered in the Spanish NHS Minimum Data Set (MDS) under category 410 of the ICD-9-CM (International Classification of Diseases, Ninth Revision, Clinical Modification). A total of 10 495 STEMI admissions was identified in the period from January 2001 to December 2013. Each of the events recorded in the MDS was reviewed by accessing computerized medical records held by the Galician Health Service (IANUS system). When necessary, paper medical records were also consulted. The coordination, verification, and supervision of data collection was the responsibility of an assigned cardiologist in each area (M. López in Ferrol, M. Santás in Lugo, and G. Aldama in A Coruña). In total, 3712 events were excluded. Exclusion criteria were a principal diagnosis other than STEMI, an old STEMI diagnosis unrelated to the hospital admission in question, duplicate events, or failure to meet the criteria for STEMI stipulated in European Society of Cardiology clinical practice guidelines and the third universal definition of myocardial infarction.11,12 The study finally included 6783 patient events.

Computerized medical records (IANUS system) or paper files were consulted to collect demographic and clinical data, treatment variables, and vital status for each patient at discharge, at 1-year of follow-up, and at the latest follow-up examination. Latest follow-up was defined as the date of the most recent entry recording the patient's vital status. Variables were entered in a database via a specially designed web-based platform allowing remote, simultaneous, and certified access.

The IPHENAMIC study defined 2 periods for comparison (figure 1), one covering the period before implementation of the PROGALIAM network (January 2001 to December 2005) and the second covering the period since the network launch (January 2006 to December 2013).

The IPHENAMIC study was approved by the Comité Autonómico de Investigación Clínica de Galicia (Registry Code 2014/257) and was entered in the ClinicalTrials.gov clinical studies registry with the identifier NCT02501070.

Discrete variables are expressed as percentages. Quantitative variables are expressed as mean±standard deviation. Comparisons between discrete variables were made by the chi-square test, and comparisons between quantitative variables were made by the Student t-test.

A survival analysis based on Cox regression was used to explore the impact of the PROGALIAM network on the study population, compare mortality between the 2 periods, and assess treatment equity between areas in both periods. The adjustment variables were those related to patient medical history, STEMI location, and clinical presentation (see “Cox regression models” in the supplementary data). The presence of collinearity in the model was tested by obtaining tolerance values and the variance inflation factor. In addition, each variable in the models was assessed to verify the proportional hazards hypothesis, as described by Grambsch and Therneau.13 When the proportionality assumption was violated, the variable was obligatorily included in the model, together with its time interaction (table 1 of the supplementary data).

For each of the models, adjusted mortality curves were obtained at 30 days, 1 year, and 5 years for the pre-PROGALIAM and PROGALIAM periods. Adjusted mortality curves were also obtained and compared for the 3 geographical areas in both periods. For all comparisons, we obtained hazard ratios and corresponding 95% confidence intervals (95%CI).

Statistical analyses were performed with IBM-SPPSS version 24.0 (IBM; Armonk, New York, United States) and R version 3.5.1. Differences were considered statistically significant at P <.05.

Mean patient age was 66 years; 76.2% were men, 54% were smokers, 48% had hypertension, 40% had hypercholesterolemia, and 21% had diabetes.

Overall differences in clinical profile and STEMI presentation between study periods are summarized in table 1. Study-period differences for each of the 3 areas are shown in table 2 of the supplementary data.

Comparisons by area revealed differences in patient profile and STEMI presentation that were especially pronounced between the A Coruña area and the Lugo and Ferrol areas (table 2).

Of the STEMI patients treated during the 2001-2013 period, 2115 did not receive reperfusion (31.2%). The 2 main causes of nonreperfusion were a >24 hour delay in contacting the health system (1166 patients [55.2%]) and medical decision (783 patients [37%]). Of the 4668 reperfused patients, 1604 (34.4%) were reperfused by thrombolysis. PPCI was given to 3064 patients (65.6%), and in 2278 of these patients (74.3%) vascular access was via the radial route. The most frequent culprit artery was the anterior descending artery (1409 patients [46%]), and half of these patients had multivessel disease. The reperfusion success rate was 98%, as defined by the presence of TIMI 3 flow and residual stenosis <20% on angiography. The median delay between symptom onset and first medical contact was 120 min, and the median delay between first medical contact and reperfusion was 142 min; median overall ischemia time was 277 min.

Comparison of the 2 study periods revealed major changes in the treatment of STEMI patients after the launch of the PROGALIAM network (table 3). The reperfusion rate, reperfusion strategy used, and unadjusted 30-day mortality are plotted against year of treatment in figure 2. The chart reveals a change in reperfusion strategy dating from the launch of the PROGALIAM network (2006), with a marked increase in the number of patients reperfused by PPCI and a corresponding decline in reperfusion by thrombolysis.

Figure 2.

Annual changes in reperfusion method and unadjusted 30-day mortality among STEMI patients. The creation of the PROGALIUM network is indicated. PPCI, primary percutaneous coronary intervention; PROGALIAM, Programa Gallego de Atención al Infarto Agudo de Miocardio; STEMI, ST-segment elevation myocardial infarction; TL, thrombolysis.

(0.29MB).

These changes are also evident from comparison of the 2 study periods in each of the 3 areas (table 3 of the supplementary data).

In-hospital complications among the 6783 patients were as follows: 547 (8.1%) had recurrent myocardial ischemia, defined as a new episode of characteristic chest pain accompanied by electrocardiogram changes; 177 (2.6%) had a repeat myocardial infarction, defined as a new episode of chest pain with or without electrocardiogram changes and accompanied by an increase in myocardial injury markers; 77 (1%) had a stroke or a transient ischemic attack; 784 (11.6%) had cardiogenic shock; and 242 (3.6%) had an actionable bleeding episode (Bleeding Academic Research Consortium types 2, 3, or 514). Median follow-up time was 68 months [interquartile range, 28-109 months]. The rates of 30-day, 1-year, and 5-year mortality were 11.9%, 16.7%, and 27.2%, respectively.

After the creation of the PROGALIAM network, there was a significant decline in the proportion of patients experiencing complications during hospitalization. The rates of 30-day, 1-year, and 5-year mortality in this period were also lower, both in the total population (table 4) and in each of the 3 study areas (table 4 of the supplementary data).

To determine whether the decline in survival was due to the implementation of the PROGALIAM network, we adjusted the mortality rates in each study period according patient clinical characteristics and STEMI location and presentation. In the PROGALIAM period, adjusted mortality was significantly lower in the total population and in each area at 30 days (figure 1 of the supplementary data), 1 year (figure 2 of the supplementary data), and 5 years (figure 3).

Figure 3.

Adjusted 5-year mortality curves for the pre-PROGALIAM period (red) and the PROGALIAM period (blue). The charts show the hazard ratios (pre-PROGALIAM vs PROGALIAM) and the corresponding confidence intervals. A) Total population. B) Population in the A Coruña area. C, Population in the Lugo area. D, Population in the Ferrol area. 95%CI, 95% confidence interval; CHUAC, Complejo Hospitalario Universitario de A Coruña; HAM, Hospital Arquitecto Marcide; HR, hazard ratio; HULA, Hospital Universitario Lucus Augusti.

(0.45MB).

To assess the success of PROGALIAM in achieving equity in treatment and outcomes for STEMI patients, we compared adjusted mortality between the different areas. Adjusted 5-year mortality curves are shown in figure 4. In the pre-PROGALIAM period, 5-year adjusted mortality was higher in the Lugo and Ferrol areas than in the A Coruña area, where the regional PPCI referral hospital is located (figure 4A). After implementation of the PROGALIUM network, these differences in mortality disappeared (figure 4B). The same result was found in the analysis of areas and study periods for mortality at 30 days (figure 3 of the supplementary data) and at 1 year (figure 4 of the supplementary data).

Figure 4.

Adjusted 5-year mortality curves for the 3 areas of northern Galicia in PROGALIAM: A Coruña (green), Lugo (red), and Ferrol (blue). The charts show hazard ratios (pre-PROGALIAM vs PROGALIAM) and the corresponding confidence intervals. A) pre-PROGALIAM period. B) PROGALIAM period. 95%CI, 95% confidence interval; HR, hazard ratio; PROGALIAM, Programa Gallego de Atención al Infarto Agudo de Miocardio.

(0.24MB).

This study has several limitations. Although we accessed all relevant STEMI patient data through an exhaustive search of the MDS and individually reviewed each event, this was an observational study, and we therefore cannot exlude bias in patient selection. Moreover, the database may not have recorded important confounding factors that therefore could not be included as adjustment variables. Similarly, the introduction of new treatments is likely to have disproportionately benefitted the more recent group. Nevertheless, the consistency in the decline in mortality for all periods examined (30 days, 1 year, and 5 years) and across all 3 geographical areas supports the conclusion that the implementation of the PROGALIAM STEMI network was responsible for the observed improvements in patient survival.