Correction: Impact of cardiac rehabilitation and treatment compliance after STsegment elevation myocardial infarction (STEMI) in France, the STOP SCA+ study

Emeline Laurent^1,2Lucile Godillon^1*Marc-Florent Tassi^1,3Pierre Marcollet⁴Stéphan Chassaing⁵Marie Decomis⁶Julien Bezin⁷Christophe Laure⁸Denis Angoulvant^10,9Grégoire Range⁸Leslie GRAMMATICO-GUILLON^11,12*

• ¹Public Health Unit, Centre Hospitalier Regional Universitaire de Tours, Tours, France
• ²Research Unit EA 7505 “Education, Ethics and Health”, University of Tours, Tours, France, Tours, France
• ³Universite de Tours Faculte de Pharmacie de Tours, Tours, France
• ⁴Cardiology, Centre Hospitalier Bourges, Bourges, France
• ⁵Cardiology, Private Hospital NCT+, Tours, France
• ⁶Cardiology, Pole Sante Oreliance, Saran, France
• ⁷INSERM, Team AHeaD, Departement Hospitalo-Universitaire de Pharmacologie de Bordeaux, Bordeaux, France
• ⁸Cardiology, Hopitaux de Chartres, Chartres, France
• ⁹Cardiology, Centre Hospitalier Regional Universitaire de Tours, Tours, France
• ¹⁰Universite de Tours Faculte de Medecine, Tours, France
• ¹¹Faculté de Médecine, Université de Tours, Tours, France
• ¹²Public Health, Centre Hospitalier Regional Universitaire de Tours, Tours, France

Compliance part in patient health outcome will need further modelling to accurately study its impact. Matching clinical and medico-administrative databases proved to be relevant for assessing outcomes at a large scale. Key learning points What is already known • Although the compliance with a cardiac treatment and cardiac rehabilitation immediately after a myocardial infarction are key factors for improving the prognosis, a significant knowledge gap regardingless is known about compliance maintenance at one year remains. What this study adds • At one year, few poor adverse outcomes occurred and were not associated with compliance to the cardiac tri-therapy, while they were associated with the absence of cardiac rehabilitation. • Matching two complementary clinical and medico-administrative databases proved to be reliable for assessing outcomes on a large scale (4,000 individuals over 5 years). Introduction Cardiovascular diseases are a major cause of death in France, mainly due to coronary artery disease and its major complications, i.e. acute coronary syndrome [1–3]. Acute ST elevation myocardial infarction (ST EMI) is a frequent common and severeerious presentation [4–6]. The reference treatment for STEMI consists in performing timely revascularization either by primary angioplasty (percutaneous intervention PCI) or by pharmacological fibrinolysis. PCI is the preferred alternative if when it can be performed in less thanwithin 120 minutes between of diagnosis and the interventional procedure [3, 7–9]. This timely intervention aims to, in order to shorten reduce the myocardial ischaemia duration and thereby improve improving the functional recovery and the survival rates [8, 10–12]. The prescription and compliance with a cardiac treatment for tertiary prevention are also key factorscrucial factors for improving thein enhancing ST EMI prognosis [4]. Optimal and sustained pharmacological management following myocardial infarction (BASI: Beta-blocker, Anti-platelet therapy, Statin, Angiotensin-converting enzyme Inhibitor) is essential for preventing ischemic recurrences and mortality. Current guidelines strongly recommend dual antiplatelet therapy (DAPT), combining aspirin and a P2Y12 inhibitor, for one year post-PCI, along with long-term LDL-cholesterol lowering therapy Indeed, drug management after myocardial infarction (BASI: Beta-blocker, Anti-platelet therapy, Statin, Angiotensin-converting enzyme Inhibitor) must be optimized and maintained over time to prevent ischaemic recurrences and death, with a high-level recommendation for a dual antiplatelet therapy DAPT (including aspirin and P2Y12 inhibitor) in the following one-year period after PCI, along with a long-term LDL-cholesterol lowering therapy [4, 13]. However, little is known about compliance to secondary prevention cardiac tri-therapy at one year, due to difficulties in real life assessment; as well as about the factors associated with this compliance and eventually, the impact of a lack of compliance. T he implementation of an automated system to monitor immediate outcomes and long-term follow-up of ST EMI patients, including medication compliance, could yield high-quality data with minimal additional cost and time investment Using an automated procedure to monitor the immediate outcomes of STEMI patients and their follow-up, including medication intake, could provide high-quality data without significant cost and time waste [14]. Therefore, in the French region of Centre-Val de Loire (approximately 2.5 million inhabitants and 6 interventional cardiology centres ICC), an electronic registry was established in 2014 (CRAC registry). This registry is fully integrated with the coronary activity report software and has automatic daily updates [3, 8]. The CRAC registry has already provided insights into several factors that impact ST EMI mortality, such as patient age, Killip score at admission, or the positive impact of calling the French medical emergency number [3, 9]. However, some registry variables related to the one-year follow-up, particularly therapeutic compliance, are collected by phone calls, which can sometimes be incomplete (not collected for 13% of the patients due lost to follow-up/death). These missing data could be collected using the National Health Insurance database (Système National des Données de Santé, SNDS). T he main primary objective of the STOP-SCA+ cohort study, STOP-SCA+, was to describe compliance to secondary prevention cardiac triple i-therapy (i.e. aspirin, P2Y12 inhibitor, and statin) oamongf patients over the one-year period after following STEMI, by matching the CRAC registry with the SNDS database. T he secondary objectives were to identify i) the factors associated with poor adverse outcomes (death or ischaemic complications) at one year, especially the impact of rehabilitation care and cardiac triple -therapy compliance; and ii) the factors associated with compliance to the secondary prevention cardiac tri-ple therapy. Methods T his study was a retrospective, multicentre and observational cohort study with a probabilistic matching (in the absence of a common direct identifier for each patient) between two databases, the CRAC registry and the medico-administrative SNDS database. Eligible patients were all patients aged 18 years or older, diagnosed with STEMI, who underwent coronary angiography or PCI within 24h of symptom onset, in five French ICC, between January 1, 2014 and December 31, 2018, as identified from the CRAC registry [3]. Exclusion criteria were the absence of consent, death occurring during the inclusion ICC hospital stay, or the absence of matching between CRAC registry and SNDS database. Data sources and matching T he CRAC Registry includes up to 150 variables for each patient, including pre-admission, hospital and follow-up data, as described elsewhere [3]. Among the 4,179 patients included in the CRAC registry over the study period, only 147 were lost to follow-up at one year (3.5%), underlying the quality of the monitoring and CRAC registry. T he SNDS includes information on the vital status, and all reimbursement data for out-of-hospitalpatient drug prescriptions and consultations, along with all hospitalisations in both public and private sectors in France (although without it does not contain detailed datadetail about drug consumptions), linked by a unique encrypted patient number [14, 15]. The pProbabilistic matching between the CRAC registry and the SNDS was performed as described detailed in the Supplementary material S1 (Tables S1.1 and S1.2). Each of the seven matching steps had demonstrated excellent performance parameters, with predictive positive values reaching >99.9% for each step (the lowest parameter performance was a sensitivity estimation at 82% for in one step involving only a few small number of cases). Overall, 96% of the CRAC registry cases could bewere successfully matched with their counterpart SNDS unique cases in the SNDS (Supplementary figure S1.3). T his study was compliant with French regulation about data protection and authorized by the French Data Protection Board (Commission Nationale de l'Informatique et des Libertés, CNIL), decision DR-2021-025 LAURENT et al. Frontiers Media SA (January 28th, 2021), as required by French regulation [16]. This study required neither information nor consent of the individuals included. Main outcome and variables of interest T he main primary outcome was defined as the occurrence of an ischaemic complication or death during within the first year after following being discharged alive from the ICC in patients who were discharged alive. The complications were extracted from the CRAC registry, both i) ischaemic event: stent thrombosis, myocardial infarction, non-haemorrhagic stroke, unplanned revascularisation, target lesion revascularisation; and ii) haemorrhagic event: haemorrhagic stroke, and severe bleeding (BARC ≥3). The vVital status was checked using both CRAC registry and SNDS data, covering the one-year period post-dischargeup to one year after discharge. For compliance assessment, three therapeutic classes were considered: aspirin, P2Y12 inhibitors, and statin. T hese medications were considered separately or combined as i) DAPT (aspirin + P2Y12 inhibitor), or ii) tri-therapy: DAPT + statin. Patients with one anti platelet agent (aspirin or P2Y12 inhibitor) associated with an anticoagulant were assimilated to DAPT patients. This association allowed to considerenabled the identification of patients with a contraindication to aspirin. For each medication, the SNDS database provides the delivery dates, which were used as proxies for the actual administration dates; and the number of boxes and tablets dispensed in at community pharmacies afteran out-of-hospital pharmacy after discharge (the prescribed duration, however, was not availableduration prescribed not known). Compliance to triple -therapy was the main variable of interest and was defined in two different ways. First, as a patient variable, by the percentage of days covered (PDC) by drug medication deliveries up tover the course of o onethe year (or up tountil death if it occurred within the before one year). For this analysis, a good compliance was defined by as a PDC ≥80% [17–19]. Second, compliance was also defined as a time-dependent variable, with a period of exposure defined as a period covered by a delivery of all three medications, whereas a non-exposure period corresponded was defined asto a period of at least one day when during which at least one of the three medications was missing, out of three. Each All hospitalisation periods, both whether forin acute care or rehabilitation care, was were considered as a full-exposure periods , provided that medication deliveries were registered either before or after the hospital stay, since in-hospital drug administrations are not recorded in the SNDSfor deliveries registered before or after the hospitalisation, as hospital deliveries are not reported in the SNDS. For cardiac rehabilitation, all hospitalisations with at least one diagnosis code Z50.0 as main care were considered, including outpatient stays. T he other variables of interest included socio-demographics characteristics (age, sex, social deprivation according to French DEPrivation index "Fdep" for each French area [20], divided in 5 quintiles), underlying conditions (such as hypertension, diabetes mellitus), medical history (such as myocardial infarction, stroke), clinical presentation at onset (including the Killip score), procedural data, left ventricular ejection fraction (LVEF) at discharge and haemorrhagic complications over the year after discharge; extracted from the CRAC registry. Cardiac Data on cardiac rehabilitation and general practitioner (GP) consultations during the year following discharge were extracted from the SNDS over the year after discharge. The reasons for the absence ofnon-admission to cardiac rehabilitation are were not available in the SNDS. Rehospitalizations were studied and defined as acute care hospitalisations with at least one overnight stay, occurring after discharge from the index stayhospitalisation. Other cardiac drugs delivered (beta-blockers, angiotensin-converting enzyme inhibitor) could not be analysed in the STOP-SCA+ study, due to variations in dosage and posology which are not captured inthat could not be tracked via the SNDS. Statistical analyses First, patients'features, clinical presentations and one-year outcomes were described in terms of numbers and percentages/frequencies. Median times between discharge from STEMI hospital stay and admission in a rehabilitation center were calculated, along with their first (Q1) and third (Q3) quartiles. Compliance was described at a patient level using the PDC. Second, the factors associated with the occurrence of an ischaemic complication or death were identified, with i) log-rank tests in univariable analyses, then ii) Cox modelling in multivariable analysis, including tri-ple therapy as time-dependent variable. Cardiac rehabilitation was also included as a time-dependant variable in this model, in order to control a potential immortal time bias. For the final model (see below), the absence of interaction between cardiac rehabilitation and compliance was checked. Hazard-ratios (HR), along with their 95% confidence interval (95%CI) were reported. We also performed sensitivity analyses to study the association between cardiac tri-ple therapy and complication/death occurrence (i) by adjustment on a propensity score (ii) and by inverse probability of treatment weighting (IPTW). T hird, after checking the absence of interaction between cardiac rehabilitation and compliance in the previous model, a secondary analysis was performed to assess factors associated with compliance (defined as the proportion of days covered PDC ≥80%) for cardiac tri-iple therapy in a subgroup of patients without major adverse cardiac or cerebral events during the subsequent year. After a univariable analysis using Chi-square tests, a multivariable logistic regression model was performed, giving odds-ratio (OR), along with their 95%CIs. For all multivariable analyses, all variables with p<0.2 in univariable analysis were included in the initial model, the final model being selected through a descending stepwise process, to select variables with p<0.05, along with clinically pertinent variables (as defined by cardiologists), sex and age. For all tests, the threshold for statistical significance was set to 5%. All analyses were conducted using SAS® Enterprise Guide software (SAS® Institute Inc., Cary, NC, USA), version available on the national SNDS portal at the time of the analyses. Results From Of the 4,179 eligible patients from the CRAC registry, 3,768 were included (90% of the eligible patients): discharged alive with available SNDS data over for the one-year follow-up year (Supplementary S1 - figure S1.3). The mean age was 62 years old [min-max 18-96], with a male-to-female ratio of approximately 3:1 with three times more men than women (table 1). Concerning the clinical presentation at onset, 3.6% had a Killip at class III or IV, the main most common infarct locationlocalisation was an inferior ischaemia (53.7%) and the pre-PCI TIMI flow was grade 0-1 in almost nearly 61% of cases. More than half of the patients were smokers (53%), 40% had hypertension, 22% reported a family history of coronary artery disease. Obesity was observed in 21% of the cases, and diabetes mellitus was present in 14%. Among the cohortOverall, 93% of the cohort underwent revascularization (84% had a primary angioplasty). At discharge, 20% of the patients had a LVEF <40% (table 1). During the one-year follow-up, 98% of the patients were prescribed at least once aspirin, 90% a P2Y12 inhibitor and 97% a statin, resulting in 89% of the patients with at least one delivery of a cardiac tri-ple therapy (n=3,362). Among these patients, 53% were compliant to their triple -therapy (PDC ≥80%) (figure 1, S2, S3 and S4). At one year, 66% of the STEMI patients had access to a cardiac rehabilitation centre (table 2). The median time between from discharge after the acute care for STEMI patient discharge from acute care setting to admission to a cardiac rehabilitation center was 8 days (Q1: 1 day - Q3: 20 days). The median time between fromthe reperfusion and to the admission in to a cardiac rehabilitation center was 14 days (Q1: 8 days-Q3: 26 days). These median times were not different differ between the patients with or without complications. Over the one-year follow-up, 303 patients (8.0%) had a pejorative adverse outcome: ischaemic complication (n=194; 5.1%) and/or death (n=130; 3.5%) (table 2). This proportion varied according to patients' characteristics and management (table 3). In multivariable analyses, compliance was not associated with complications over the year (HR 1.16 [0.86-1.57]), while the absence of cardiac rehabilitation (2.31 [1.73-3.08]) was associated, as well as female sex (1.54 [1.08-2.19]), renal impairment (2.87 [1.49-5.53]) initial clinical presentation (pejorative high Killip 2.04 [1.19-3.50], and LVEF <40% at discharge 2.22 [1.65-2.99]) (figure 2, S5 and S7). No interaction was found between cardiac rehabilitation and cardiac tri-ple therapy compliance. Social deprivation was not associated with pejorative adverse outcomes in univariable analysis (results not reported), thus not included in the multivariable analysis. Studying the factors associated with cardiac tri-therapy good compliance to cardiac triple therapy (PDC ≥80%), participation in cardiac rehabilitation was a contributive significant positive factor (OR=1.55 [1.34-1.79]) (figure 3, S6), along with a younger age, being a male sex and, the absence of a medical history of stroke or myocardial infarction (figure 3). Discussion Main findings T his study showed that few complications nor deaths occurred in the year following a STEMI. Only 8% of the patients had experienced poor outcomes, lower than reported elsewhere, although direct comparisons were are limited due to different differences in inclusion criteria and outcome definitionss [21–23]. Pejorative Adverse outcomes were mainly related to patients' pre-existing conditions (age ≥65 years old, female sex, renal impairment and diabetes), the initial severity of the STEMI (Killip 3-4 and LVEF at discharge <40%), but also associated with the absence of cardiac rehabilitation (adjusted HR 2.3), whereas the effect of compliance to cardiac tri-iple therapy could not be demonstrated through the sensitivity analyses. Our results aligned are consistent with those of the FAST-MI registry which showed reported a significant reduction in one-year mortality risk among French patients admitted to in a French cohort of patients admitted to cardiac rehabilitation after an acute myocardial infarction [24, 25]. Here, we found that almost all STEMI patients were prescribed at least once aspirin (98% of patients), a P2Y12 inhibitor (90%) or a statin (97%), resulting in 89% with at least one triple-therapy prescription, but only one half of them had a PDC≥80%. This definition of compliance was very stringent, however, the initial results showed that the absence of medicine uptake once in a while had no pejorative adverse consequences. It might have been interesting to consider different thresholds of compliance in sensitivity analysis. We also emphasized the importance of considering medical compliance both as a static parameter, such as the PDC over a defined period, and as a dynamic parameter, such as a time-dependent variable, regardless of the total duration of medication deliveries. Our results provided insights in a field where no consensus currently exists regarding the optimal duration of DAPT or statin therapy following STEMI. According to the 2017 ESC guidelines, "multiple studies have shown that shortening DAPT to 6 months, compared with 12 months or longer, reduces the risk of major bleeding complications, with no apparent trade-off in ischaemic events", while "two major studies have shown the benefit towards reduction of non-fatal ischaemic events in patients receiving longer than 12 months of DAPT". The guidelines conclude that "no formal recommendations are possible for the use of clopidogrel or prasugrel beyond 1 year"[26]. Regarding the long-term prescription of statins, although strongly recommended, the ESC guidelines do not specify a duration. Instead, they state that "lipids should be re-evaluated 4–6 weeks after the [acute coronary syndrome] to determine whether the target levels have been reached and regarding safety issues; the lipid lowering therapy can then be adjusted accordingly." Long term follow-up of patients enrolled in clinical trials testing lipid-lowering therapy in high-risk patients supports a favorable risk-beneficit ratio for prolonged LDL-lowering therapy [27]. Other post-STEMI treatments recommended in the BASI were not considered for this study [4, 13], due to numerous posology adaptations over time, not allowing to reliably assess compliance. Furthermore, our study took into account data on post-STEMI cardiac rehabilitation through from the SNDS database, which appeared toidentifying it as be a major factor to associated with improved the one-year outcomes, as previouslyconsistent with previous findings suggested, although a strong robust causal model remains has yet to be built [28, 29]. Our results were are particularly encouraging, asapproximately two-thirds of the patients benefited participated infrom cardiac rehabilitation, which was higher than previous prior reportsdata, which indicate that only about one-third of coronary patients in Europe and 25% in the USA receive any form of cardiac rehabilitation where about a third of coronary patients in Europe receives any form of cardiac rehabilitation and only a 25% in the USA [19, 24]. However, it was recently shown that patients with ST EMI were admitted to cardiac rehabilitation twice as often as patients without ST-segment elevation [30]. In France, Puymirat et al demonstrated the benefits of cardiac rehabilitation after admission for acute myocardial infarction with a risk reduction of one-year mortality [24]. Study limitations First, observational design could not provide such causal inferences as a randomised control trial (RCT) [31]. A way of emulating a RCT is to perform a propensity score-based study. However, some studies highlighted that using a propensity score does not change the odds ratios as compared to logistic or Cox regression models [32, 33]. For overcoming imbalances in baseline characteristics, we performed sensitivity analyses, including inverse probability of treatment weighting (IPTW), giving discordant results, not allowing to demonstrate an association between compliance and pejorative negative outcomes. It would be interesting to repeat the study in a few years when a higher number of events is reached. Moreover, we could not eliminate a potential immortal time bias, even if we showed that the median time from discharge to rehabilitation admission was not different between patients with or without any pejorative adverse outcomes. Besides, the definition of compliance through the SNDS medico-administrative database could be challenging [18, 34]. T he SNDS contains the dispensations and not the administration, and delivery discontinuations cannot be identified, resulting in a possible misestimation of compliance; under-estimation for patients for whom the delivery was stopped due to health or tolerance issues, whereas overestimation of compliance in case of self-discontinuation by patients who do not feel the usefulness of the treatment in the absence of symptoms (treatment delivered but not taken). Moreover, the SNDS did not allow to take into account a number of other confounders that could affect both medication compliance and outcomes, such as environmental factors. Social deprivation was considered through the French DEPrivation index "Fdep" [20] and no interaction was found with cardiac tri-therapy compliance. Given these limitations, the inconclusive results findings about regarding compliance and pejorative adverse outcomes could did not allow any for definitive conclusions. Further studies researchare is needed to drop the concernsaddress these concerns, and larger populations that could be available in real-world registries, such as France PCI, the national extension of the CRAC registry, could provide valuable insights. T he CRAC registry has also some limitations: the inclusion criterion is limited to patients admitted in ICC. T hus, ST EMI patients without any invasive procedure, even if rare, are not analysed, conducing to a selection bias. However, previous studies estimated that hospitalized STEMI patients without coronary angiography corresponded to less than 5% in France [19, 35]. Strengths T he originality and interest of our study are multiple. First, our study was the first to date matching two complementary databases: the clinico-biological data from the CRAC registry (subgroup of France-PCI), including a large STEMI population of nearly 4,000 individuals over 5 years, and the medico-administrative data of the SNDS, allowing to increase the completeness and accuracy of the results. The SNDS proved its epidemiological interest, allowing to catch complete and large-scale data, especially after discharge in outpatient care, which usually require an onsite follow-up led by local research technicians, impacting cost and resulting in an underestimation of the results due to lost to follow-up/dead patients. Moreover, one of the major strengths of this database is its ability to facilitate analyses based on real-world data, allowing the study of the entire care continuumwas the possibility to perform analyses based on real-life data allowing to study the real care-continuum. The regional registry, initially covering 5 ICC, has is now extended expanded to include over to over 89 ICC across the countrynationwide as part of the in the France PCI initiative. As Since the SNDS is part of the French Health data hub (www.francepci.fr), this will enable the extension of our findings to a national level scale in a near future. EventuallyFurthermore, the matching process between datasets was both reliable and efficient and could provide an estimation estimate of the operational resources required, assessed in real-life settings. T ogether, these factors support the generalizability and robustness of our findings.All these elements assess the generalizability and robustness of the findings. Conclusions T his study showed that patients admitted for emergency revascularization at the acute phase of a STEMI mainly had favourable outcomes at one year, with few complications or death. Pejorative Adverse outcomes appeared to be related to patients' characteristics, initial clinical presentation, and access to rehabilitation care, whereas no impact could be observed for non-compliance. Matching two complementary clinical and medico-administrative databases proved to be reliable for assessing outcomes on a large scale. This permanent and multicentric database with systematic long-term follow-up in ICC provides excellent quality of data and immediate feedbacks at a large scale, without high cost nor changing usual practice due to a fully integrated electronic medical report. This large reliable database could soon provide care quality improvement linked to benchmarking and promote robust studies. Tables Table 1. Patients' characteristics and hospital management – The STOP-SCA+ study Patients included n % missing Total 3,768 100 Age, years (mean [min-max]) 62 [18-96] 0 Women 936 24.8 0 Comorbidities Overweight (25 ≤ BMI < 30kg/m2) 1,599 42.5 4 Obesity (BMI ≥ 30kg/m2) 789 21.0 4 Renal impairment 52 1.4 36 High blood pressure 1,510 40.3 25 Diabetes 519 13.9 25 Smoking (current or past) 2,006 53.5 18 Medical history T CA/myocardial infarction/coronary bypass 465 12.3 0 Stroke 94 2.5 3 Peripheral vascular disease 99 2.6 10 Family history of coronary disease 797 21.8 115 Killip at the admission 72 III 52 1.4 IV 80 2.2 Ischaemia location 31 Inferior 2,006 53.7 Anterior 1,448 38.7 Lateral 268 7.2 Circumferential 15 0.4 Reperfusion procedure 3,497 92.8 0 Primary angioplasty 3,171 84.2 Isolated fibrinolysis 54 1.4 Angioplasty after fibrinolysis 272 7.2 Isolated coronary angiography 271 7.2 Pre-PCI TIMI flow grade 0-1 2,246 60.8 71 Post-PCI TIMI flow grade 0-1 or no reflow 77 2.4 519* Median time (min) from symptom onset to ECG 100 803** from ECG to primary angioplasty 101 653** from ECG to fibrinolysis 20 73** LVEF at discharge < 40% 19.5 362 BMI: Body Mass Index; TCA: Transluminal Coronary Angioplasty PCI : Percutaneous Coronary Intervention; LVEF: Left Ventricular Ejection Fraction; * Including patients without any PCI (n=263) ** before 2015 median time not measured and counting as missing data Table 2. Patients' outcome at one year after STEMI – The STOP-SCA+ study T he STOP-SCA+ study: 1-year outcome Patients included n % Total 3,768 100 One-year re-hospitalisations Acute care (full hospitalisation) 1,817 48.2 with vascular disease 1,327 35.2 with emergency admission for vascular disease 299 7.9 Rehabilitation care In-hospital stay (≥1 night) 1,787 47.4 One-day hospitalisation (0 night) 994 26.4 With cardiac rehabilitation* 2,478 65.8 Ischaemic complication** or death 303 8.0 Ischaemic complication 194 5.1 Stent thrombosis 21 0.6 Myocardial infarction 50 1.3 Non-haemorrhagic stroke 17 0.5 Unplanned revascularisation 157 4.2 T arget lesion revascularisation 32 0.8 Death 3.5 Haemorragic complication 73 1.9 * ICD-10 diagnosis code Z50.0 "Cardiac rehabilitation" as the main care registered ** stent thrombosis, myocardial infarction, non-haemorragic stroke, unplanned revascularisation, target lesion revascularisation Table 3. Ischaemic complication and/or death at one year after STEMI according to their characteristics and management – The STOP-SCA+ study Total Among whom, ischaemic complication and/or death (n=303) Univariable analysis p-value (log-rank) n % (in line) All patients 3,768 8.0 Age ≥65 years old 1,601 10.7 <0.01 Women 936 7.7 0.67 Comorbidities Obesity (BMI ≥ 30kg/m2) 789 7.2 0.34 Renal impairment 52 26.9 <0.01 High blood pressure 1,510 10.1 <0.01 Diabetes 519 14.6 <0.01 Smoking (current or past) 2,006 7.7 0.33 Medical history T CA/myocardial infarction/coronary bypass 465 12.9 <0.01 Stroke 94 13.8 0.03 Peripheral vascular disease 99 19.2 <0.01 Family history of coronary disease 797 7.5 0.56 Killip = 3 or 4 132 17.4 <0.01 Reperfusion procedure 0.83 Primary angioplasty 3,171 8.1 LAURENT et al. Frontiers Media SA Isolated fibrinolysis 11.1 Secondary angioplasty 272 7.7 Isolated coronary angiography 271 7.4 LVEF at discharge <40%* 665 12.2 <0.01 No cardiac rehabilitation 1,290 12.3 <0.01 * missing data: n=362 BMI: Body Mass Index; LVEF: Left Ventricular Ejection Fraction; MI: Myocardial Infarction; TCA: Transluminal Coronary Angioplasty Figure titles Figure 1. Cardiac drug compliance at one year after STEMI – the STOP-SCA+ study. PDC: proportion of days covered by the studied medication Figure 2. Factors associated with an ischaemic complication and/or death at one year after STEMI – the ST OP-SCA+ study. LVEF: Left Ventricular Ejection Fraction Figure 3. Factors associated with compliance for the cardiac tri-therapy (PDC ≥80%) at one year after ST EMI – the STOP-SCA+ study TCA: Transluminal Coronary Angioplasty ; GP: General Practicioner; Figure 2. Factors associated with an ischaemic complication and/or death at one year after STEMI – the ST OP-SCA+ study. LVEF: Left Ventricular Ejection Fraction There was a mistake in figure 3 as published. It was reversed with figure 2. The corrected figure 3 appears below. Figure 3. Factors associated with compliance for the cardiac tri-therapy (PDC ≥80%) at one year after ST EMI – the STOP-SCA+ study TCA: Transluminal Coronary Angioplasty ; GP: General Practicioner; The original version of this article has been updated.