The aims of FAST-MI 2010 were to provide an extensive description of the population of patients admitted for AMI throughout the French territory, to determine whether differences in terms of population characteristics existed across regions, to assess the management of the patients suffering from AMI, and to determine the implementation of practice guidelines in a real world setting. Other objectives were to assess the correlations between management strategies and in-hospital outcomes, to determine the correlations between genetic polymorphisms and morbidity-mortality in relation with the effects of medications, and to determine the relationships between biomarkers and morbidity-mortality after MI. Also, FAST-MI will enable historic comparisons with the previous French cohorts constituted since 1995 and will document the long-term (up to 10 years) outcomes of patients presenting with AMI at the end of 2010, in relation with early and long-term management.
Organisation and funding
Type of study
Prospective, multi-centre observational study in continental France and Corsica. FAST-MI 2010 is a registry of the French Society of Cardiology, supported by the Société Française de Médecine d'Urgence, SAMU de France, the Collège National des Cardiologues des Hôpitaux and the Collège National des Cardiologues Français.
Funding and data propriety
The study is sponsored by the French Society of Cardiology and funded by unrestricted grants of the following companies: MSD, and AstraZeneca, the Daiichi-Sankyo-Eli-Lilly alliance, Glaxo-Smith-Kline, Novartis, and sanofi-aventis. Complementary grants will be sought for dedicated research projects within the main study.
Conduct of the registry and legal issues
The cardiologists who participated in the registry were not supposed to modify their therapeutic approach in any way. In the centres participating in the bio-bank, however, an additional 10–60 ml blood sample was collected at the time the routine blood sample was drawn.
Written informed consent was provided by each patient for participating in the study. The data from patients dying early (ie, before informed consent was obtained), were collected and recorded in the database, unless opposed to by the patient's next of kin when informed.
The study was conducted in compliance with Good Clinical Practice, French law and the French data protection law. The protocol was reviewed and approved by the Committee for the Protection of Human Subjects (CPP) of Saint Louis University Hospital Paris Ile de France IV. The data recorded and the way they are handled and stored were reviewed and approved by the Comité consultatif sur le traitement de l'information en matière de recherche dans le domaine de la santé (CCTIRS) and the Commission Nationale Informatique et Liberté (CNIL). Clinicaltrials.gov identifier: NCT01237418.
Each patient had to meet the following criteria:
- 1) Man or woman aged 18 years or over.
- 2) Admitted within 48 h of symptom onset to an intensive care unit (ICU) or a Cardiology department for an AMI characterised by elevation of troponins or CK-MB associated with at least one of the following:
- – symptoms compatible with myocardial ischaemia,
- – development of new abnormal Q waves,
- – ST-T changes compatible with myocardial ischaemia (ST segment elevation or depression, T wave inversion);
- 3) Having agreed to take part into the study.
ST-segment elevation myocardial infarction (STEMI) was diagnosed when ST elevation ≥1 mm was seen in at least two contiguous leads in any location on the index or qualifying ECG, or when presumed new left bundle branch block or documented new Q waves were observed.
In the absence of ST-segment elevation, patients meeting the inclusion criteria were considered to have NSTEMI.
Patients who died very early after admission and for whom cardiac markers were not measured were included if they had compatible signs or symptoms associated with typical, unequivocal ST changes. Patients admitted after resuscitation of a cardiac arrest were included only if the cardiac arrest had been preceded by chest pain suggestive of AMI, having justified a call to emergency services.
Patients participating in clinical trials could be included in the registry, and information on trial participation was recorded.
Exclusion criteria were: AMI with symptom onset >48 h from first call, unstable angina, iatrogenic AMI immediately following cardiovascular procedures (patients with late stent thrombosis, defined as stent thrombosis occurring after hospital discharge could be included), and diagnosis of AMI refuted in favour of alternative diagnoses, such as acute myocarditis.
A list of all intensive care or coronary care units admitting patients with AMI in academic institutions, general hospitals, army hospitals and private clinics was established at the beginning of 2010. The physicians in charge were then asked to participate in the study. In all, 279 centres were listed, 224 of which initially accepted to participate in the study and 213 actively participated, including one general hospital which received no AMI patient during the 1-month study period. Participation rate was 76% and the centres were distributed across the whole country ().
Geographical distribution of the centres participating in FAST-MI 2010.
In addition, approximately 100 further centres could receive AMI patients at their emergency rooms, but referred these patients to other institutions. These centres were not asked to participate in the registry, as the patients were included at the ‘receiving’ institutions.
The participating centres comprised: 111 general hospitals, 49 private clinics, 13 private not-for-profit clinics, 38 academic hospitals, and 2 army hospitals.
Patients were recruited consecutively from ICU/cardiology departments over a period of one month (from 1 October 2010). A physician was responsible for study recruitment in each centre. A list of all patients admitted within 48 h of symptom onset for a suspicion of ACS was established at each institution, and inclusion/exclusion criteria were checked. A computerised case record form (CRF) was filled-in for each eligible patient, based on the hospital records and additional specific questionnaires. Data were recorded on-line by dedicated research technicians from a contract research organisation (CRO) (ICTA, Fontaine-lès-Dijon, France) who went to each centre on a weekly basis. In the case of incomplete data from the source patient files, the research technicians contacted the local investigator to obtain the missing information.
Recruitment could extend up to 2 months in the centres willing to do so. The population included during the first month by the totality of participating centres is used as illustrative of current management and outcomes on the scale of the whole country, while the whole population will be used to answer specific medical interrogations on a larger sample.
What is recorded?
Cardiovascular and non-cardiovascular medical history, risk factors, and clinical course, including symptoms, admission and worst Killip class, therapeutic management in the pre-hospital setting, during the first 48 h, during the hospital stay and at discharge, were recorded. Left ventricular ejection fraction, when assessed at entry and at any time during the hospital stay was recorded. Routine laboratory results on admission were collected. All in-hospital complications were recorded. In all, the electronic CRF comprised 385 items, to which was added information on medications administered before the index event, pre-hospital, during the first 24 and 48 h and at hospital discharge; for oral medications, doses were recorded.
In 131 centres, a biology substudy collected blood and serum samples for the assessment of DNA, RNA (120 Centres) and centralised measurement of biological markers (77 Centres).
Follow-up data are collected by research technicians based at the French Society of Cardiology (SFC) under the supervision of the Unité de Recherche Clinique (URC-EST) of Assistance Publique-Hôpitaux de Paris (APHP), using the following sequential procedure, which was implemented for the follow-up of patients participating in the FAST-MI 2005 registry: (1) consulting the registry offices of the patients' birthplaces for death certificates; (2) contacting the patients' general practitioners and/or cardiologists; and (3) contacting the patients or their relatives. In many instances, written contact is followed by telephone interviews with the patients or their family.
For each reported event leading to hospitalisation or death, hospital discharge reports are sought and analysed by at least one physician from the research team. All cases of cardiovascular events are centrally reviewed by at least one physician. We will specifically collect information on the occurrence of acute coronary syndromes, myocardial revascularisation, hospitalisation for heart failure, stroke, bleeding events requiring hospitalisation, hospitalisation for cancer. Cases in which the final diagnosis appears unclear or debatable are reviewed by a three-member critical events committee.
The quality of the data was insured by automated data checks when the electronic CRF were filled-in, and by the fact that all the data entered were verified, and modified if necessary, by external research assistants. This methodology had already been used for the FAST-MI 2005 survey with research assistants from the same CRO and an external audit in three of 21 regions showed data concordance in >90% of the cases.
Completeness of the data was adequate for most variables (eg, missing values <4% for height and weight; <2% for admission blood pressure and heart rate; <0.2% for risk factors or cardiovascular history; GRACE score calculated in 94.5%).
Once entered into the electronic CRF, data were stored in a central database at the SFC, Paris. Data management is ensured in conjunction between the SFC and the cardiology departments of Toulouse and HEGP university hospitals.
Use of the data and cross-linkage
Data are available for research to any participating clinician submitting an analysis plan to the executive committee (fastmi2010/at/yahoo.fr). In addition, any centre can claim access to its own data.
The database will be cross-linked with the French Institute for National Statistics and the French national death database (CepiDC) to gather additional information on the cause of death.
Finally, to evaluate changes in practice in France over the past fifteen years, individual data from the 1995, 2000, 2005 and 2010 registries will be merged in a common database.
For quantitative variables, means, standard deviations, and minimum and maximum values are calculated. In addition, medians with the IQR are calculated for some variables. Discrete variables are presented as percentages. Comparisons are made with chi-square or Fisher' s exact tests for discrete variables, and by unpaired T tests, Mann–Whitney U tests, Wilcoxon sign-rank tests or one-way analyses of variance for continuous variables. Odds and hazard ratios are given with their 95% CIs.
Multivariate analyses of predictors of short-term outcome (in-hospital, and 30 days) will use multiple logistic regression analysis.
Long-term survival curves will be estimated using the Kaplan Meier method and comparisons will be made using log-rank tests. Independent correlates of survival will be determined using a multivariate Cox model. Variables included in the final multivariate models will comprise those with a significance level <0.15 in the univariate analyses, unless otherwise stated. For specific analyses regarding management options, propensity scores will be calculated using multiple logistic regression analyses and propensity-score-matched cohorts will be constructed.
For genetic analyses, all single-nucleotide polymorphisms evaluated will be tested for deviation from Hardy–Weinberg equilibrium with the use of χ2 tests.
Statistical analyses are performed at Université Paris Descartes (Nicolas Danchin, MD, PhD), Université of Toulouse (Jean Ferrières, MD, PhD, Vincent Bataille, PhD), and Université Pierre et Marie Curie Paris 06 (supervised by Tabassome Simon, MD, PhD), using the SAS V.8.2 (SAS Institute), STATA (V.9), NCSS, or IBM-SPSS V.20.0 (IBM) softwares.
Strengths and weaknesses
The main strength of the present registry is that detailed data were recorded to characterise the patients, their management and outcomes. Overall more than 385 fields had to be filled-in, not counting the totality of the medications used (including doses for oral medications). The percentage of institutions participating in the registry was high. We could not include, however, centres from overseas French territories, so that our data are illustrative of Metropolitan France only.
Though the cardiologists were asked not to modify their therapeutic approach, we have no way to make sure that participating in the registry had absolutely no impact on practices. As we committed not to divulge the data of individual centres to anyone beside the clinicians in the centres themselves, however, there was no incentive for them to change their habits.
Contrary to mandatory registries, FAST-MI was based upon the voluntary participation of the centres. It therefore cannot pretend to represent an exhaustive collection of AMI patients during the study period. Of note, most centres not participating were low-volume centres, so that our data are likely skewed towards management practices that are mostly used in larger-volume centres. Major bias, however, is unlikely. In a population of AMI patients collected from medico-administrative databases during the first 6 months of 2006, mean age was 65 years12
; in 2010, national data based on the ICD I21, I22 and I23 codes found a mean age of 68.0 years and 33% women (data from the Institut National de Veille Sanitaire). Overall, it is likely that some patients, (eg, very elderly) were hospitalised outside cardiology departments, or may have been admitted to smaller hospitals and therefore not included in our registry. The clinical outcomes in FAST-MI must be interpreted in this context. Temporal comparisons with previous French surveys using the same methodology, however, should not be affected by this potential bias, which was common to all previous surveys.