BOTH PRIMARY PERCUTANEOUS

Transcription

1 ORIGINAL CONTRIBUTION Association Between Timeliness of Reperfusion Therapy and Clinical Outcomes in ST-Elevation Myocardial Infarction Laurie Lambert, PhD Kevin Brown, MSc Eli Segal, MD James Brophy, MD, PhD Josep Rodes-Cabau, MD Peter Bogaty, MD See also pp 2156 and Context Guidelines emphasize the importance of rapid reperfusion of patients with ST-elevation myocardial infarction (STEMI) and specify a maximum delay of 30 minutes for fibrinolysis and 90 minutes for primary percutaneous coronary intervention (PPCI). However, randomized trials and selective registries are limited in their ability to assess the effect of timeliness of reperfusion on outcomes in real-world STEMI patients. Objectives To obtain a complete interregional portrait of contemporary STEMI care and to investigate timeliness of reperfusion and outcomes. Design, Setting, and Patients Systematic evaluation of STEMI care for 6 months during in 80 hospitals that treated more than 95% of patients with acute myocardial infarction in the province of Quebec, Canada (population, 7.8 million). Main Outcome Measures Death at 30 days and at 1 year and the combined end point of death or hospital readmission for acute myocardial infarction or congestive heart failure at 1 year by linkage to Quebec s medicoadministrative databases. Results Of 1832 patients treated with reperfusion, 392 (21.4%) received fibrinolysis and 1440 (78.6%) received PPCI. Fibrinolysis was untimely ( 30 minutes) in 54% and PPCI was untimely ( 90 minutes) in 68%. Death or readmission for acute myocardial infarction or heart failure at 1 year occurred in 13.5% of fibrinolysis patients and 13.6% of PPCI patients. When the 2 treatment groups were combined, patients treated outside of recommended delays had an adjusted higher risk of death at 30 days (6.6% vs 3.3%; odds ratio [OR], 2.14; 95% confidence interval [CI], ) and a statistically nonsignificant increase in risk of death at 1 year (9.3% vs 5.2%; OR, 1.61; 95% CI, ) compared with patients who received timely treatment. Patients treated outside of recommended delays also had an adjusted higher risk for the combined outcome of death or hospital readmission for congestive heart failure or acute myocardial infarction at 1 year (15.0% vs 9.2%; OR, 1.57; 95% CI, ). At the regional level, after adjustment, each 10% increase in patients treated within the recommended time was associated with a decrease in the region-level odds of overall 30-day mortality (OR, 0.80; 95% CI, ). Conclusion Among patients in Quebec with STEMI, reperfusion delivered outside guideline-recommend delays was associated with significantly increased 30-day mortality, a statistically nonsignificant increase in 1-year mortality, and significantly increased risk of the composite of mortality or readmission for acute myocardial infarction or heart failure at 1 year. JAMA. 2010;303(21): Author Affiliations: Agence d évaluation des technologies et des modes d interventions en santé (Drs Lambert and Bogaty and Mr Brown), Jewish General Hospital, McGill University/Urgences-santé (Dr Segal), and McGill University Health Center, McGill University (Dr Brophy), Montreal, Québec, Canada; and Institut universitaire de cardiologie et pneumologie de Québec, BOTH PRIMARY PERCUTANEOUS coronaryintervention(ppci)and fibrinolysis are well-recognized treatmentsforst-segmentelevation myocardial infarction(stemi) in international guidelines, and benefits are maximizedwhentreatmentoccursearly. 1,2 Although meta-analyses of randomized clinical trials support the superiority of PPCIoverfibrinolysis, 3 thegeneralizabilityoftheseresultstoeverydayclinicalpractice may be compromised by the highly selective recruitment of both patients and centers in these trials and the inability to assess the effect of timeliness of treatment onoutcomesbecauseofthegenerallyrapid performance of PPCI. 3,4 In STEMI registries, the observed delays for PPCI often largely exceed those realized in clinical trials, 5-7 and a substantial proportion of patients do not receive either PPCI or fibrinolysis within maximum delays recommendedininternationalguidelines(90 minutes for PPCI and 30 minutes for fibrinolysis). 1,2 Intheseregistries,shorter treatment delays are associated with betteroutcomes 5,8-10 andresultsofthe2treatments tend to be similar. 11,12 However, registry data also have limitations, including selection bias, information bias (noninclusion of patients referred for PPCI but not receiving it), incomplete and inappropriately short follow-up, frequent exclusion of patients transferred for PPCI, and residual confounding (lack of detailed measurements of important covariates). 7,8,10,13,14 An assessment of bias within a large patient registry found that a substantial pro- Québec City, Canada (Drs Rodes-Cabau and Bogaty). Corresponding Author: Laurie Lambert, PhD, Cardiology Evaluation Unit, Agence d évaluation des technologies et des modes d interventions en santé (Quebec Healthcare Assessment Agency), 2021 Union Ave, Montreal, QC H3A 2S9, Canada 2148 JAMA, June 2, 2010 Vol 303, No. 21 (Reprinted with Corrections) 2010 American Medical Association. All rights reserved.

2 portion of eligible patients were not enrolled and that these patients received poorer-quality care and had 3-fold higher in-hospital mortality than enrolled patients. 13,14 Finally, registries based on voluntary participation of institutions usually provide information from wellperforming, motivated academic and research-oriented centers but tend to miss an important proportion of patients in the larger real-world setting. To address these issues, the Agence d évaluation des techniques et modes d interventions en santé, a governmentsupported but independent health care evaluation agency in Quebec, Canada, recently undertook a systematic provincewide evaluation of STEMI care in Quebec (population, 7.8 million). The objectives were to obtain a contemporary portrait of reperfusion treatments and their delays across Quebec and to determine whether STEMI reperfusion treatment outside of the guidelinerecommended delays is associated with poorer outcomes than treatment within recommended delays. METHODS The evaluation covered a 6-month period (October1,2006,toMarch31,2007)and included all acute care Quebec hospitals (n=80) that treated at least 30 acute myocardial infarctions (AMIs) in the preceding year. These 80 hospitals treat more than 95% of all AMIs occurring in Quebec and are distributed across 15 of Quebec s18administrativehealthcareregions. Some regions with small populations and similar systems of care were combined in our analyses (n=12 regions). Ethics approval was obtained from each of the 80 hospitals as well as by the provincial ethics board (Quebec Commission for Access to Information), which waived the need for patient consent because this evaluation involved no intervention or patient contact and anonymity was ensured. Data Collection and Patient Identification In each hospital, a certified medical record librarian was designated, received an individualized on-site training session, and followed a standardized data collection process to abstract all information directly from medical charts and enter data onto a secure, centralized Web site with automatic encryption. At the completion of data entry, date-time values were independently validated by senior medical librarians in a representative 10% sample and concordance was very high (median difference for door to reperfusion therapy=0 min [interquartile range {IQR}, 0-0]). 15 Study eligibility was assessed via an algorithm in all patients with appropriate presenting symptoms, a final diagnosis ofami(internationalclassificationofdiseases, NinthRevisioncode410), andahospital discharge date between October 1, 2006, andmarch31, Patientswhose AMIoccurredaftertheirinitialemergency department presentation were excluded. ForeachpotentialSTEMIpatient, thefirst electrocardiogram (ECG) taken after arrival at the emergency department was sent to the coordinating center, where it was interpreted by 2 cardiologists and an emergency physician. Patients included in the evaluation had to have 1 of the following features: (1) received fibrinolysis in the 4 hours following triage at the firstemergencydepartment;(2) weresent to a catheterization laboratory in the 4 hours following triage at the first emergency department and had mention of STEMI inthemedicalchartor,ifnot,evidenceofstemionthefirstecgvalidated at the coordinating center; or (3) did not receive fibrinolysis and were not sent to acatheterizationlaboratoryinthe4hours after triage at the first emergency department but had mention of STEMI in the medical chart and evidence of STEMI on the first ECG validated at the coordinating center. In data analyses, all patients who satisfied criteria 1 and 2 as described were considered to have received reperfusion treatment (intention to treat). However, patients fulfilling criteria 2 who were sent for PPCI but who did not undergo PPCI were not evaluated in analyses of timeliness of treatment. Patients fulfilling criteria 3 were considered as not having received reperfusion treatment (no intention to treat). For patients included more than once during the observation period, only the first event was retained in the analysis. Variables and Outcomes Clinical factors included age, sex, earliest in-hospital measure of systolic blood pressure, heart rate at first in-hospital ECG, and symptom duration. A TIMI index was calculated for each patient: (heart rate [age/10] 2 )/systolic blood pressure. 16 The TIMI index was used in statistical analyses rather than the 3 factors individually because it has been validated as a powerful predictor of mortality in STEMI patients and improved the statistical power of the multivariate model. 16 Anterior STEMI and left bundle branch block were centrally identified based on the admission ECG. Treatment time was calculated from patient arrival at the first hospital (door) to either start of fibrinolysis (needle) or first dilation (balloon). To ascertain patient comorbidities, the recorded principal diagnosis as well as the 15 secondary diagnosesofallhospitaladmissionsinthe 5 years preceding the date of admission for the index STEMI were identified in Quebec s administrative hospital discharge database. We included diagnoses in the index STEMI admission only for comorbidities considered to be chronic to avoid confusing an index admission complication with a comorbidity. The primary end points were death at 30 days, death at 1 year, and the combined end point of death or hospital readmission for congestive heart failure or AMI at 1 year. Other end points included PCI at 1 year (excluding the index admission) and coronary artery bypass graft surgery at 1 year. All end points were determined by linkage with provincial medicoadministrative databases whose reliability has been shown. 17,18 Statistical Analysis Mortality at 30 days and at 1 year was ascertained according to type of treatment (fibrinolysisorppci), treatmentdelayand timeliness ( 30 minutes for fibrinolysis and 90 minutes for PPCI), clinical factors, and comorbidities. After adjustment for age, no significant interaction was 2010 American Medical Association. All rights reserved. (Reprinted with Corrections) JAMA, June 2, 2010 Vol 303, No

3 observed between any clinical risk factor or comorbidity and type of treatment. Thus, the odds ratio (OR) and 95% confidence interval (CI) for the association between each variable and mortality were calculated with the 2 treatment groups combined. Statistical significance was based on 2-sided testing with P.05. To measure the association between timeliness of reperfusion (fibrinolysis and PPCI combined) and mortality, we used a multivariate logistic regression model that included timeliness of treatment, 3 clinical factors (TIMI index, sex, and anterior MI), 4 comorbidities (congestive heart failure, renal failure, peripheral vascular disease, and cancer), and 2 confounding factors (arrival by ambulance and transfer status) that have been shown to be associated with both timelinessoftreatmentandmortality. 8,19,20 The number of variables in the model was limited to have a reasonable ratio of factors to outcome events. The choice of variables was made a priori on the basis of consensual clinical judgment, established associations with mortality, 6,10,11 and sufficient prevalence within the study population ( 5%). All 10 variables were included in each multivariate model. Given the number of STEMI patients identified during the observation period, we had 80% power to detect an OR of 2.0 for the effect of timeliness on 30-day mortality and an OR of 1.8 for 1-year mortality. We had 80% power to detect an OR of 1.6 for the combined outcome. To address concerns that low-risk patients might be preferentially selected for reperfusion treatment, an analysis of 30- daymortalityattheregionallevelwasperformed that included all STEMI patients (with and without intention to treat). Overall regional 30-day mortality was plotted as a function of the regional proportion of patients treated within maximum recommended delays as well as a function of the proportion of patients treatedwithppci. Anunadjustedinversevariance weighted least squares regression was performed to ascertain the slope of each association. A multilevel logistic regression analysis was conducted that included the 10 variables as fixed effects and the intercept term as a region-level randomeffect. 21 Ofthe10variables,those referring to process of care were included as region-level variables (percentage of patientsreceivingtimelyreperfusion, who used an ambulance, or who were transferred for PPCI) while the remaining 7 riskfactorswereincludedattheindividual patient level. Because of the inclusion of patients without intention to treat in this model, anadditionalregion-levelvariable was added to the model representing the percentageofpatientswhoreceivedreperfusion treatment. The model estimated the region-level intercepts and their overall variance. Due to the possibility of biased multivariate effect estimates in nonexperimental studies when the ratio of outcomes to potential confounders is low, a matched propensity score analysis was used as a sensitivity analysis of the effect of treatment timeliness. 21,22 The propensity score was calculated by using 18 variables (age, female sex, systolic blood pressure 100 mm Hg, heart rate 100/ min, transfer status, mode of patient arrival, previous MI, previous chronic heart failure, cardiac dysrhythmia, peripheral vascular disease, chronic obstructive pulmonary disease, diabetes, hypertension, cerebrovascular disease, cancer, renal failure, previous hospital admission, and presence of 2 comorbidities) in a multivariate logistic regression for timeliness of treatment. Patients receiving timely treatment were matched 1:1 with patients receiving late treatment if they had a propensity score within 0.1 of each other and in such a way that minimized total absolute differences in propensity scores. 23 Match balance was assessed by standardized and absolute differences on each covariate. 24 Conditional logistic regression was used to assess the OR and 95% CI of death (at 30 days and at 1 year) and the combined outcome for the matched pairs. In another sensitivity analysis, we examined timeliness and mortality by type of treatment for patients who, on the basis of the first ECG, were unanimously considered by the emergency physician and the 2 cardiologists at the coordinating center to have an unequivocal STEMI. In a third sensitivity analysis, we compared mortality by type of treatment for patients who were treated in study hospitals where there was an exclusive treatment practice (hospitals treating 95% of STEMI patients with either fibrinolysis or PPCI). In these hospitals, the choice of reperfusion treatment is unlikely to have any association with patient characteristics. Statistical analyses were performed using SAS software 25 and R software 26 as well as the R add-ons lme4 27 and Matching. 28 RESULTS Reperfusion Treatment Profile We identified 6734 patients who presented to an emergency department with both characteristic symptoms and a final diagnosis of AMI, of whom 2361 (35%) had STEMI. Five (0.2%) were excluded because no link could be established with the other medicoadministrative databases, leaving 2356 patients. The majority of STEMI patients were treated with PPCI (1440/2356 [61.1%]); 16.6% (392/2356 patients) were treated with fibrinolysis (tenecteplase in 99%) and 22.2% (524/2356 patients) neither had fibrinolysis nor were sent for PPCI within 4 hours of emergency department arrival. Thus, of the patients treated with acute reperfusion (n=1832), 78.6% (n=1440) were sent for PPCI and 21.4% (n=392) received fibrinolytic therapy. Of the patients sent for PPCI, 34.7% (n=499) were direct admissions to a PPCI center while 65.3% (n=941) were transferred from a non-pci center. Of patients treated with fibrinolysis, 79.1% (n=310) were eventually sent for cardiac catheterization, the majority (53%) more than 12 hours later. Among patients referred for PPCI, 10.6% (n=153) did not undergo the procedure for various reasons, such as resolution of symptoms, a nonsignificantly narrowedculpritlesion,uncertaintyabout theculpritlesion,inabilitytopasstheguide wire, or a decision to perform coronary artery bypass graft surgery instead. Clinical Characteristics Patient characteristics and comorbidities bytypeoftreatmentareshownintable JAMA, June 2, 2010 Vol 303, No. 21 (Reprinted with Corrections) 2010 American Medical Association. All rights reserved.

4 The median age was 58 years in patients who received fibrinolysis and 60 years in PPCI patients(p=.01). The median TIMI indexwas2pointshigherinppcipatients than in patients who received fibrinolysis(p=.003).comparedwithpatientssent forppci,morefibrinolytic-treatedpatients presented to the emergency department within 3 hours of the start of symptoms whilefewerofthemarrivedbyambulance. They were less likely than patients sent for PPCI to have a history of arterial hypertension or diabetes but more likely to have a history of congestive heart failure. Timeliness of Reperfusion Among patients who underwent PPCI, the median door-to-balloon time was 110 minutes (IQR, minutes); 32% had a delay within the maximum recommended limit of 90 minutes. Among patients who were directly admitted to a PPCI center, the median door-toballoon time was 83 minutes (IQR, minutes); 57% were treated within 90 minutes. For transferred patients, the median door-to-balloon time was 123 minutes (IQR, minutes), and only 19% were treated within 90 minutes. For patients who received fibrinolysis, the median delay was 33 minutes (IQR, minutes); 46% had a door-toneedle time within the maximum recommended delay of 30 minutes. We examined timeliness of treatment in patients who had an admission ECG that was considered unequivocal for STEMI. These patients constituted 52% (669/1287) of PPCI patients and 62% (242/392) of fibrinolysis patients. Timeliness of treatment of PPCI-treated patients directly admitted to a PPCI center increased to 70% (from 57%); for patients transferred for PPCI, it increased to 25% (from 19%); and for patients treated with fibrinolysis, timeliness of treatment increased from 46% to 55%. The 2 other factors most strongly associated with untimely reperfusion treatment in univariate analyses were transfer from a non-pci center to a PPCI center (OR, 4.63; 95% CI, ) and arrival without use of an ambulance (OR, 1.92; 95% CI, ). Outcomes Among patients who neither received fibrinolysis nor were sent for PPCI within 4 hours of triage, all-cause mortality was 18%at30daysand29%at1year.TABLE2 shows that all-cause mortality in the 2 treatment groups was similar. At 30 days, 6.1% of patients who received fibrinolysisand5.6%ofpatientssentforppcihad died (OR, 0.91; 95% CI, ). At 1year, 7.4% whoreceivedfibrinolysisand 8.3% of patients sent for PPCI had died (OR, 1.13; 95% CI, ). Whenonly Table 1. Patient Characteristics According to Type of Treatment a Characteristics patients with an admission ECG unequivocal for STEMI were considered, mortality at 30 days was 4.5% for those receiving fibrinolysis and 4.7% in those sent for PPCI (OR, 1.06; 95% CI, ). At 1 year, mortality was 5.4% for those receiving fibrinolysis and 7.3% in those sent for PPCI (OR, 1.51; 95% CI; ). In patients who actually underwent PPCI, mortality was 5.2% at 30 days and 7.9% at 1 year compared with 9.2% and 11.7%, respectively, in those who did not undergo the procedure. Fibrinolysis (n = 392) PPCI (n = 1440) P Value Demographic and clinical factors Age, median (IQR), y 58 (51-68) 60 (52-71).01 TIMI index, median (IQR) b 17.3 ( ) 19.3 ( ).003 Female 83 (21.2) 374 (26.0).06 Heart rate 100/min 39 (9.9) 158 (11.0).63 Systolic blood pressure 100 mm Hg 33 (8.4) 156 (10.8).19 Anterior myocardial infarction 128 (32.6) 460 (31.9).84 Left bundle-branch block 8 (2.0) 31 (2.2).95 Symptom duration to triage 3 h 321 (81.9) 983 (68.3).001 Triage during work hours 211 (53.8) 727 (50.5).26 Transfer prior to reperfusion (65.3).001 Arrival by ambulance 223 (56.9) 905 (62.8).04 Comorbidities Previous myocardial infarction 47 (12.0) 195 (13.5).47 Previous PCI 18 (4.6) 107 (7.4).06 Previous heart failure 18 (4.6) 32 (2.2).02 Peripheral vascular disease 23 (5.9) 93 (6.5).76 Chronic obstructive pulmonary disease 38 (9.7) 153 (10.6).66 Diabetes 49 (12.5) 249 (17.3).03 Renal disease 23 (5.9) 91 (6.3).83 Hypertension 137 (34.9) 598 (41.5).02 Cerebrovascular disease 11 (2.8) 50 (3.5).62 Cancer 16 (4.1) 57 (4.0).97 Abbreviations: IQR, interquartile range; PCI, percutaneous coronary intervention; PPCI, primary percutaneous coronary intervention. a Data are expressed as No. (%) unless otherwise indicated. b TIMI index is calculated as (heart rate [age/10]2 )/systolic blood pressure. Table 2. Adverse Outcomes by Type of Treatment Outcomes Fibrinolysis (n = 392) No. (%) PPCI a (n = 1440) PPCI vs Fibrinolysis, Unadjusted OR (95% CI) Death at 30 d 24 (6.1) 81 (5.6) 0.91 ( ) Death at 1 y 29 (7.4) 120 (8.3) 1.13 ( ) Readmission for heart failure at 1 y 11 (2.8) 50 (3.5) 1.22 ( ) Readmission for AMI at 1 y 14 (3.6) 36 (2.5) 0.69 ( ) Death or readmission for heart failure or AMI at 1 y 53 (13.5) 196 (13.6) 1.01 ( ) PCI at 1 y (post index hospitalization) 25 (6.4) 78 (5.4) 0.84 ( ) Coronary artery bypass graft surgery at 1 y 46 (11.7) 114 (7.9) 0.65 ( ) Abbreviations: AMI, acute myocardial infarction; CI, confidence interval; OR, odds ratio; PPCI, primary percutaneous coronary intervention. a Includes all patients sent for PPCI whether or not PPCI was performed American Medical Association. All rights reserved. (Reprinted with Corrections) JAMA, June 2, 2010 Vol 303, No

5 About two-thirds of patients received reperfusion therapy in hospitals where choice of reperfusion treatment was exclusive or nearly so ( 95% either PPCI orfibrinolysis).inexclusivetreatmenthospitals, 30-day mortality in patients who received fibrinolysis was 5.8% compared with 5.0% in patients sent for PPCI (OR, 0.85; 95% CI, ). At 1 year, mortality was 6.8% for patients who received fibrinolysis and 7.3% for patients sent for PPCI (OR, 1.08; 95% CI, ). Incidence of the combined end point (death or readmission for heart failure or AMI) at 1 year was 13.5% for fibrinolysis patients and 13.6% for PPCI patients (OR, 1.01; 95% CI, ) (Table 2). Occurrence at 1 year of coronary artery bypassgraftsurgerywaslowerintheppci groupthaninthefibrinolysisgroup(7.9% vs 11.7%, respectively; OR, 0.65; 95% CI, ), while occurrence of PCI (after the index hospitalization) was similar in the 2 groups (5.4% vs 6.4%, respectively; OR, 0.84; 95% CI, ). Table 3. Adverse Outcomes by Type and Timeliness of Treatment Outcomes No. (%) of Patients OR (95% CI) a 30 min (n = 182) Fibrinolysis 30 min (n = 210) Mortality at 30 d 6 (3.3) 18 (8.6) 2.75 ( ) Mortality at 1 y 8 (4.4) 21 (10.0) 2.41 ( ) Readmission for AMI at 1 y 7 (3.3) 7 (3.8) 0.86 ( ) Readmission for CHF at 1 y 2 (1.1) 9 (4.3) 4.02 ( ) Mortality, CHF, or AMI at 1 y 16 (8.8) 37 (17.6) 2.22 ( ) PPCI b 90 min (n = 417) 90 min (n = 870) Mortality at 30 d 14 (3.4) 53 (6.1) 1.87 ( ) Mortality at 1 y 23 (5.5) 79 (9.1) 1.71 ( ) Readmission for AMI at 1 y 8 (1.9) 21 (2.4) 1.26 ( ) Readmission for CHF at 1 y 8 (1.9) 33 (3.8) 2.02 ( ) Mortality, CHF, or AMI at 1 y 39 (9.4) 125 (14.4) 1.63 ( ) Combined Fibrinolysis and PPCI b Recommended Delay (n = 599) Recommended Delay (n = 1080) Mortality at 30 d 20 (3.3) 71 (6.6) 2.04 ( ) 2.14 ( ) c Mortality at 1 y 31 (5.2) 100 (9.3) 1.87 ( ) 1.61 ( ) c Mortality, CHF, or AMI at 1 y 55 (9.2) 162 (15.0) 1.75 ( ) 1.57 ( ) c Abbreviations: AMI, acute myocardial infarction; CHF, congestive heart failure; CI, confidence interval; OR, odds ratio; PPCI, primary percutaneous coronary intervention. a Odds ratios are unadjusted unless otherwise indicated. b Patients sent to the catheterization laboratory but who did not receive PPCI were excluded from these analyses. c Adjusted OR. Adverse Outcomes and Timeliness of Treatment In an unadjusted analysis, for each treatment type, mortality was significantly higher in patients treated outside of recommended times than in patients treated within recommended times (TABLE 3). At 30 days, patients treated with PPCI outside of recommended times had higher mortality (OR, 1.87; 95% CI, ) as did patients treated outside of recommended times with fibrinolysis (OR, 2.75; 95% CI, ). The increase in mortality of untimely treatment was sustained at 1 year for both untimely PPCI (OR, 1.71; 95% CI, ) and untimely fibrinolysis (OR, 2.41; 95% CI, ). Within each treatment group, there was nonsignificantly less readmission for congestive heart failure with timely compared with untimely treatment (Table 3). Overall, the composite end point of death or readmission for congestive heart failure or AMI at 1 year occurred significantly more frequently when treatment was untimely among both patients who received fibrinolysis and those who underwent PPCI (Table 3). Untimely fibrinolysis was associated with a worse prognosis than timely PPCI, just as untimely PPCI was associated with a worse prognosis than timely fibrinolysis. When the 2 treatment groups were combined, the risk of adverse events at 1 year was higher (OR, 1.75; 95% CI, ) in patients who were not treated within maximum recommended delays. When treatment delay beyond maximum recommended guidelines was examined in 30-minute intervals in patients who underwent PPCI, there was an increasing risk of 30-day mortality with increasing treatment delay for both patients admitted directly to a PPCI center (P=.003) and patients transferred for PPCI (P.001). Association Between Patient Characteristics and Mortality Univariate associations with 30-day mortality are shown in TABLE 4. Clinical factors significantly associated with increased 30-day mortality were female sex and higher TIMI index. Each comorbidity was associated with a statistically significant increase in risk of death. Arrival by ambulance was associated with an increased risk of death, while being transferred for PPCI was associated with a lower risk of death that was not statistically significant. Association Between Timeliness of Treatment and Outcomes: Multivariate Analysis After multivariate adjustment, estimates of the association between timeliness of treatment and adverse events remained largely unchanged from unadjusted estimates (Table 3). For mortality in the 2 treatment groups combined, the adjusted relative risk associated with untimely treatment was 2.14 (95% CI, ) at 30 days and 1.61 (95% CI, ) at 1 year. For the combined outcome of death or readmission for heart failure or AMI at 1 year, 2152 JAMA, June 2, 2010 Vol 303, No. 21 (Reprinted with Corrections) 2010 American Medical Association. All rights reserved.

6 the adjusted relative risk for untimely treatment was 1.57 (95% CI, ). In the propensity analysis, of the 599 patients receiving timely treatment (Table 3), a matched patient receiving late treatment was found for 504 patients (84%; difference of 5% of standard deviation for all variables). The resultant ORs from the matched propensity analysis for 30-day mortality (OR, 2.00; 95% CI, ), for 1-year mortality (OR, 1.59; 95% CI, ), and for the combined outcome (OR, 1.57; 95% CI, ) were not substantially different than those yielded by the multivariate logistic regression models. Timeliness of Reperfusion Treatment and Outcomes by Region Across Quebec health care regions, overall timeliness of treatment (PPCI and fibrinolysis combined) ranged from 19% to 76% and use of fibrinolysis and PPCI each varied from 0% to 100% (FIGURE). In the unadjusted least squares regression analysis, each 10% increase in patients treated within recommended time was associated with a 1.09% decrease (95% CI, 1.76% to 0.42%) in region-level mortality at 30 days (Figure, A). However, there was no relationship between regional use of PPCI and region-level overall mortality at 30 days (slope, ; 95% CI, to 0.070) (Figure, B). After adjustment for patient and regional characteristics, the proportion of patients treated within recommended time at the region level was significantly associated with the region s odds of death for all STEMI patients. Each 10% increase in regional timeliness was associated with a region-level odds of overall 30-day mortality of 0.80 (95% CI, ). The other region-level variables in the multilevel model were not significant. Regional mortality in patients who did not receive reperfusion treatment was not higher in regions where mortality for patients receiving reperfusion treatment was low (slope, 0.54; 95% CI, 1.39 to 2.46). This suggests that the relationship between timeliness of reperfusion and mortality was not due to selective treatment of healthier patients. Table 4. Univariate and Multivariate Results Univariate Mortality at 30 d COMMENT In this evaluation, we sought to obtain a complete contemporary portrait of STEMI reperfusion treatment in Quebec and to ascertain the effect of untimely treatment on patient outcomes. The principal findings are that (1) PPCI is the predominant reperfusion strategy in the province of Quebec, with most PPCI patients transferred from non- PCI hospitals; (2) delays to reperfusion treatment exceeded maximum recommended delays in a substantial proportion of patients treated with fibrinolysis and PPCI, particularly in patients transferred for PPCI; (3) risk of adverse events was similar in patients treated with PPCI and fibrinolysis but was higher in those treated outside of the maximum recommended delays, regardless of the reperfusion strategy; and (4) timeliness of treatment with either reperfusion mode was a strong predictor of overall regional mortality but no association was observed with choice of reperfusion treatment within health care regions. Adjusted a OR (95% CI) Mortality at 1 y, Adjusted b Mortality, CHF, or AMI at 1 y, Adjusted c Female 2.58 ( ) 1.82 ( ) 1.77 ( ) 1.47 ( ) TIMI index 1.08 ( ) 1.07 ( ) 1.06 ( ) 1.05 ( ) Anterior AMI 1.47 ( ) 1.51 ( ) 1.46 ( ) 1.82 ( ) History of CHF 4.86 ( ) 2.25 ( ) 2.99 ( ) 3.84 ( ) History of renal failure 5.11 ( ) 1.49 ( ) 2.27 ( ) 2.27 ( ) History of peripheral 3.62 ( ) 1.81 ( ) 1.64 ( ) 1.57 ( ) vascular disease History of cancer 3.25 ( ) 1.34 ( ) 2.62 ( ) 1.82 ( ) Arrival by ambulance 3.53 ( ) 2.03 ( ) 2.05 ( ) 1.21 ( ) Transfer 0.71 ( ) 0.54 ( ) 0.84 ( ) 0.83 ( ) Untimely treatment 2.04 ( ) 2.14 ( ) 1.61 ( ) 1.57 ( ) Abbreviations: AMI, acute myocardial infarction; CHF, congestive heart failure; CI, confidence interval; OR, odds ratio. a C statistic for multivariate model of 30-day mortality=0.86. b C statistic for multivariate model of 1-year mortality=0.85. c C statistic for multivariate model of combined 1-year outcome=0.78. Figure. Regional 30-Day Mortality of All STEMI Patients by Intraregional Proportion of Patients Treated Within Recommended Delays and Percentage of Patients Receiving Reperfusion With PPCI Cumulative Incidence of Mortality at 30 d, % A Patients treated within recommended delay y = ; r = 0.75 (95% CI, 0.92 to 0.31) Patients Treated Within Recommended Delay, % Cumulative Incidence of Mortality at 30 d, % B Patients treated with PPCI y = ; r = 0.14 (95% CI, 0.47 to 0.66) Patients Treated With PPCI, % Each circle represents a region or combined region. The size of the circles is proportional to the number of patients with ST-elevation myocardial infarction (STEMI) treated with reperfusion within the region. CI indicates confidence interval; PPCI, primary percutaneous coronary intervention American Medical Association. All rights reserved. (Reprinted with Corrections) JAMA, June 2, 2010 Vol 303, No

7 Our study, while consistent with registry and clinical data associating longer treatment delays with poorer outcomes, 8-10 is novel and robust in several ways. Above all, it represents not a sampling but more than 95% of all STEMI patients within a large and complex system of care and provides very recent information that transcends the relative selectivity of randomized clinical trials and most registries. Importantly, patients transferred for PPCI, patients in whom there was no initial intention to treat with reperfusion, and patients who were sent for reperfusion treatment but who did not receive it for various reasons were all included. Patients with STEMI were systematically identified on the basis of an algorithm and central interpretation of the admission ECG. Linkage with medicoadministrative databases permitted assessment of comorbidities and ensured completeness of follow-up beyond the short term not only of mortality but also of the important end points of readmission for AMI or heart failure and revascularization procedures. Sensitivity analyses, notably restricted to patients with an ECG unequivocal for STEMI and centers performing exclusive reperfusion treatments, as well as propensity analysis and region-level analysis, strengthen confidence in the observed associations. Moreover, since this study was performed in a single-payer, universal access system, observedassociationsareunlikely to be confounded by patient differences in insurance and socioeconomic status. Thus, we believe this evaluation represents a needed contribution to the evidence base for deriving clinical practice guidelines 29 andanimportantadvancein knowledge of the outcomes associated with contemporary processes of STEMI care. This real-world information is relevant both clinically and from a perspective of evidence-based health care policy and planning, 14 pointing to the lifesaving potential for approaches that focus on offering the most timely reperfusion treatment to patients with STEMI. Time, rather than mode of reperfusion, emerges as a critical determinant of outcome in this systematic evaluation of STEMI care. Regardless of reperfusion strategy, patients treated beyond maximum recommended delays had increased mortality. Occurrence of congestive heart failure, an important determinant of poor prognosis, was also increased and is consistent with the notion that longer treatment times lead to less myocardial salvage, making subsequent heart failure more likely. In contrast, longer treatment delay was not associated with an increased risk of readmission for AMI, which may not be unexpected since there is no plausible pathophysiological link between longer treatment delay and risk of new MI. The association between longer delay to reperfusion treatment and poorer outcome may be confounded because patients who are sicker may be treated later rather than earlier. For example, hemodynamically unstable patients could require more time to be stabilized before being able to undergo reperfusion treatment. Also, patients unlikely to survive a long delay to reperfusion may be less likely to be transferred for PPCI and might receive no reperfusion treatment. However, after multivariateadjusted analyses that included adjustment for clinical risk factors (TIMI index, female sex) and comorbidities (history of heart failure, peripheral vascular disease, renal failure, or cancer) as well as factors known to be associated with both timeliness and mortality (presentation by ambulance and transfer status), 8,19,20 an increase in adverse outcomes in patients with longer delays to reperfusion was observed for mortality in the short term and for a combined measure of morbidity and mortality in the long term. Moreover, a propensity analysis that matched an array of variables in patients with untimely treatment to patients with timely treatment resulted in similar observations of increased risk. Thus, this province-wide evaluation provides up-todate, real-world evidence for the emphasis expressed in international guidelines on the importance of considering timeliness of treatment when choosing the mode of reperfusion. 1,2 Faced with a choice between a timely reperfusion option and an untimely one, the former option appears associated with better patient outcomes. Consistent with results in treated patients only, we found an important and statistically significant relation between regions timeliness of treatment and survival in an analysis of all patients with STEMI. Regions that delivered reperfusion treatment beyond maximum recommended delays had higher mortality than regions that delivered reperfusion treatment within recommended delays while mode of reperfusion within the region was not associated with mortality. This regional analysis of all STEMI patients suggests that the observed benefits of timely treatment are not due to a selection bias of treating healthier patients or treating them more quickly. These results may reassure clinicians and health care policy planners in regions where PPCI is not an option that timely fibrinolysis is a very acceptable standard of STEMI care. Although retrospective collection of data has limitations, it has the important advantage of allowing a more complete and systematic process for STEMI patient identification, thus minimizing some biases. 13,14 The accuracy of data entry concerning the process of STEMI care was also found to be high 15 but this observational study has limitations. Nonrecorded data might have led to residual confounding that could have influenced our findings. However, adjustment for available confounding variables, including important clinical factors, did not materially change observed results. Although medicoadministrative databases ensured completeness of follow-up for the primary end points, they did not contain appropriate information to assess hemorrhagic events. Among patients in Quebec with STEMI, reperfusion delivered outside of guideline-recommended treatment delays was associated with significantly increased 30-day mortality, a statistically nonsignificant increase in 1-year mortality, and significantly increased risk of the composite outcome of mortality or readmission for 2154 JAMA, June 2, 2010 Vol 303, No. 21 (Reprinted with Corrections) 2010 American Medical Association. All rights reserved.

8 AMI or heart failure at 1 year. The study findings suggest that time to reperfusion rather than treatment strategy may be more important in terms of outcomes and can help inform clinical decision making to optimize care for patients with AMI presenting to hospitals in Quebec. Author Contributions: Drs Lambert and Bogaty and Mr Brown had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study conception and design: Lambert, Bogaty. Acquisition of data: Lambert, Bogaty, Segal. Analysis and interpretation of data : Brown, Lambert, Bogaty, Segal, Brophy, Rodes-Cabau. Drafting of the manuscript: Lambert, Bogaty, Brown. Critical revision of the manuscript for important intellectual content: Segal, Brophy, Rodes-Cabau. Statistical analysis: Brown, Lambert, Bogaty, Brophy, Segal, Rodes-Cabau. Obtained funding: Lambert, Bogaty. Administrative, technical, or material support: Lambert, Bogaty, Brown, Segal, Brophy, Rodes-Cabau. Study supervision: Lambert, Bogaty. Financial Disclosures: Dr Rodes-Cabau reports that he is recipient of a research grant from Boston Scientific and lecture fees from Abbott, Cordis, and Boston Scientific. No other disclosures were reported. Funding/Support: This evaluation was funded by the Ministry of Health of the province of Quebec. Role of the Sponsor: The funder had no role in the design and conduct of the study, analysis and interpretation of the data, or preparation, review, or approval of the manuscript. Additional Contributions: We dedicate this article to the memory of Konrad Jamrozik, MD, PhD. We gratefully acknowledge the important contributions of Céline Carroll, BSc, and Maude Giguère, AMA, Agence d évaluation des technologies et des modes d interventions en santé (study conception and design, acquisition of data, administrative, technical, and material support, and training of 80 designated medical record librarians). We also gratefully acknowledge other important contributions of the following persons: Luce Boyer, RN, Centre de recherche de l institut universitaire de cardiologie et de pneumologie de Québec (conception and design); Peter Faris, PhD, Alberta Bone and Joint Institute (statistical advice on propensity analyses and statistical power calculation); Jean E. Morin, MD, McGill University Health Centre (conception and design); Jean-Marie Moutquin, MD, Agence d évaluation des technologies et des modes d interventions en santé (critical revision of manuscript); James Nasmith, MD, St Paul s Hospital (analysis of data); Serge Simard, MSc, Centre de recherche de l institut universitaire de cardiologie et de pneumologie de Québec (statistical advice and analysis); Benedict Stuber-Gaumond, BSc, Centre de recherche de l institut universitaire de cardiologie et de pneumologie de Québec (creation of Web site and corresponding database); Rémy Thériault, PhD, Centre de recherche de l institut universitaire de cardiologie et de pneumologie de Québec (creation of Web site and corresponding database); and Jack V. Tu, MD, PhD, Institute for Clinical Evaluative Sciences (critical revision of manuscript). Ms Boyer, Drs Morin, Moutquin, Nasmith, and Thériault, and Messrs Simard and Stuber-Gaumond were compensated for their contributions to the project. We also thank the designated medical record librarians in the 80 participating hospitals across Quebec for acquisition of data and the Executive Committee and Scientific Committee of the Quebec Tertiary Cardiology Network (our clinical partners in this endeavor). REFERENCES 1. Van de Werf F, Bax J, Betriu A, et al; ESC Committee for Practice Guidelines. Management of acute myocardial infarction in patients presenting with persistent ST-segment elevation: the Task Force on the Management of ST-Segment Elevation Acute Myocardial Infarction of the European Society of Cardiology. Eur Heart J. 2008;29(23): Kushner FG, Hand M, Smith SC Jr, et al; American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines Focused updates: ACC/AHA guidelines for the management of patients with ST-elevation myocardial infarction (updating the 2004 guideline and 2007 focused update) and ACC/AHA/SCAI guidelines on percutaneous coronary intervention (updating the 2005 guideline and 2007 focused update): a report of the American College of Cardiology Foundation/ American Heart Association Task Force on Practice Guidelines. Circulation. 2009;120(22): Huynh T, Perron S, O Loughlin J, et al. Comparison of primary percutaneous coronary intervention and fibrinolytic therapy in ST-segment-elevation myocardial infarction: Bayesian hierarchical meta-analyses of randomized controlled trials and observational studies. Circulation. 2009;119(24): Steg PG, Lopez-Sendon J, Lopez de Sa E, et al; GRACE Investigators. External validity of clinical trials in acute myocardial infarction. Arch Intern Med. 2007; 167(1): Nallamothu BK, Wang Y, Magid DJ, et al; National Registry of Myocardial Infarction Investigators. Relation between hospital specialization with primary percutaneous coronary intervention and clinical outcomes in ST-segment elevation myocardial infarction: National Registry of Myocardial Infarction-4 analysis. Circulation. 2006;113(2): Mehta RH, Bufalino VJ, Pan W, et al; American Heart Association Get With the Guidelines Investigators. Achieving rapid reperfusion with primary percutaneous coronary intervention remains a challenge: insights from American Heart Association s Get With the Guidelines program. Am Heart J. 2008;155(6): Eagle KA, Nallamothu BK, Mehta RH, et al; Global Registry of Acute Coronary Events (GRACE) Investigators. Trends in acute reperfusion therapy for ST-segment elevation myocardial infarction from 1999 to 2006: we are getting better but we have got a long way to go. Eur Heart J. 2008;29(5): Gibson CM, Pride YB, Frederick PD, et al. Trends in reperfusion strategies, door-to-needle and doorto-balloon times, and in-hospital mortality among patients with ST-segment elevation myocardial infarction enrolled in the National Registry of Myocardial Infarction from 1990 to Am Heart J. 2008; 156(6): Nallamothu B, Fox KA, Kennelly BM, et al; GRACE Investigators. Relationship of treatment delays and mortality in patients undergoing fibrinolysis and primary percutaneous coronary intervention: the Global Registry of Acute Coronary Events. Heart. 2007; 93(12): Rathore SS, Curtis JP, Chen J, et al; National Cardiovascular Data Registry. Association of door-toballoon time and mortality in patients admitted to hospital with ST elevation myocardial infarction: national cohort study. BMJ. 2009;338:b Danchin N, Coste P, Ferrieres J, et al; FAST-MI Investigators. Comparison of thrombolysis followed by broad use of percutaneous coronary intervention with primary percutaneous coronary intervention for ST-segment-elevation acute myocardial infarction: data from the French Registry on Acute ST-Elevation Myocardial Infarction (FAST-MI). Circulation. 2008; 118(3): Kalla K, Christ G, Karnik R, et al; Vienna STEMI Registry Group. Implementation of guidelines improves the standard of care: the Viennese Registry on Reperfusion Strategies in ST-Elevation Myocardial Infarction (Vienna STEMI Registry). Circulation. 2006; 113(20): Ferreira-González I, Marsal JR, Mitjavila F, et al. Patient registries of acute coronary syndrome: assessing or biasing the clinical real world data? Circ Cardiovasc Qual Outcomes. 2009;2(6): Krumholz HM. Registries and selection bias: the need for accountability. Circ Cardiovasc Qual Outcomes. 2009;2(6): Lambert LJ, Carroll-Bilodeau C, Giguère M, et al. Reliability of process measures of care of patients presenting with ST-elevation myocardial infarction (STEMI) abstracted by hospital health record librarians: results of a field evaluation in Quebec, Canada [abstract 175]. Circ Cardiovasc Qual Outcomes. 2009; 2(3):e Wiviott SD, Morrow DA, Frederick PD, et al. Performance of the Thrombolysis in Myocardial Infarction risk index in the National Registry of Myocardial Infarction-3 and -4: a simple index that predicts mortality in ST-segment elevation myocardial infarction. J Am Coll Cardiol. 2004;44(4): Levy AR, Tamblyn RM, Fitchett D, McLeod PJ, Hanley JA. Coding accuracy of hospital discharge data for elderly survivors of myocardial infarction. Can J Cardiol. 1999;15(11): Thibault N. Suivi et cycle des naissances et des décès: qu en est-il au Québec? Québec, QC: Institut de la Statistique du Québec; _Ecostat2007M03F02.pdf. Accessed April 16, So DY, Ha AC, Turek MA, et al. Comparison of mortality patterns in patients with ST-elevation myocardial infarction arriving by emergency medical services vs self-transport (from the prospective Ottawa Hospital STEMI Registry). Am J Cardiol. 2006; 97(4): Westfall JM, Kiefe CI, Weissman NW, et al. Does interhospital transfer improve outcome of acute myocardial infarction? a propensity score analysis from the Cardiovascular Cooperative Project. BMC Cardiovasc Disord. 2008;8: Stukel TA, Fisher ES, Wennberg DE, Alter DA, Gottlieb DJ, Vermeulen MJ. Analysis of observational studies in the presence of treatment selection bias: effects of invasive cardiac management on AMI survival using propensity score and instrumental variable methods. JAMA. 2007;297(3): Cepeda MS, Boston R, Farrar JT, Strom BL. Comparison of logistic regression vs propensity score when the number of events is low and there are multiple confounders. Am J Epidemiol. 2003;158(3): Rosenbaum PR. Optimal matching for observational studies. J Am Stat Assoc. 1989;84(408): Austin PC. Assessing balance in measured baseline covariates when using many-to-one matching on the propensity-score. Pharmacoepidemiol Drug Saf. 2008;17(12): SAS Institute Inc. SAS Version Cary, NC: SAS Institute Inc; R Development Core Team. R: a language and environment for statistical computing: reference index, version Vienna, Austria: R Foundation for Statistical Computing; Accessed April 16, Bates D, Maechler M. lme4: LINEAR mixed effects models using S4 classes (R package version ) /packages/lme4/index.html. Accessed April 16, Sekhon JS. Matching: multivariate and propensity score matching with balance optimization (R package version 4.7-6). Berkeley: University of California; /Matching/index.html. Accessed April 16, Tricoci P, Allen JM, Kramer JM, Califf RM, Smith SC Jr. Scientific evidence underlying the ACC/AHA clinical practice guidelines. JAMA. 2009;301(8): American Medical Association. All rights reserved. (Reprinted with Corrections) JAMA, June 2, 2010 Vol 303, No