Association Between Timeliness of Reperfusion Therapy and Clinical Outcomes in ST-Elevation Myocardial Infarction

ORIGINAL CONTRIBUTION

Laurie Lambert, PhD Context Guidelines emphasize the importance of rapid reperfusion of patients with Kevin Brown, MSc ST-elevation myocardial infarction (STEMI) and specify a maximum delay of 30 min- Eli Segal, MD utes for fibrinolysis and 90 minutes for primary percutaneous coronary intervention (PPCI). However, randomized trials and selective registries are limited in their ability to assess James Brophy, MD, PhD the effect of timeliness of reperfusion on outcomes in real-world STEMI patients. Josep Rodes-Cabau, MD Objectives To obtain a complete interregional portrait of contemporary STEMI care Peter Bogaty, MD and to investigate timeliness of reperfusion and outcomes. Design, Setting, and Patients Systematic evaluation of STEMI care for 6 months OTH PRIMARY PERCUTANEOUS during 2006-2007 in 80 hospitals that treated more than 95% of patients with acute coronaryintervention(PPCI)and myocardial infarction in the province of Quebec, Canada (population, 7.8 million). fibrinolysis are well-recognized treatmentsforST-segmenteleva- 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 tionB myocardial infarction (STEMI) in in- heart failure at 1 year by linkage to Quebec’s medicoadministrative databases. ternational guidelines, and benefits are 1,2 Results Of 1832 patients treated with reperfusion, 392 (21.4%) received fibrinoly- maximizedwhentreatmentoccursearly. Ͼ Although meta-analyses of randomized sis and 1440 (78.6%) received PPCI. Fibrinolysis was untimely ( 30 minutes) in 54% and PPCI was untimely (Ͼ90 minutes) in 68%. Death or readmission for acute myo- clinical trials support the superiority of cardial infarction or heart failure at 1 year occurred in 13.5% of fibrinolysis patients 3 PPCI over fibrinolysis, the generalizabil- and 13.6% of PPCI patients. When the 2 treatment groups were combined, patients ityoftheseresultstoeverydayclinicalprac- treated outside of recommended delays had an adjusted higher risk of death at 30 tice may be compromised by the highly days (6.6% vs 3.3%; odds ratio [OR], 2.14; 95% confidence interval [CI], 1.21-3.93) selective recruitment of both patients and and a statistically nonsignificant increase in risk of death at 1 year (9.3% vs 5.2%; centers in these trials and the inability to OR, 1.61; 95% CI, 1.00-2.66) compared with patients who received timely treat- assess the effect of timeliness of treatment ment. Patients treated outside of recommended delays also had an adjusted higher onoutcomesbecauseofthegenerallyrapid 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, performance of PPCI.3,4 In STEMI regis- 1.08-2.30). At the regional level, after adjustment, each 10% increase in patients treated tries, the observed delays for PPCI often within the recommended time was associated with a decrease in the region-level odds largely exceed those realized in clinical of overall 30-day mortality (OR, 0.80; 95% CI, 0.65-0.98). trials,5-7 and a substantial proportion of Conclusion Among patients in Quebec with STEMI, reperfusion delivered outside patients do not receive either PPCI or fi- guideline-recommend delays was associated with significantly increased 30-day mor- brinolysis within maximum delays rec- tality, a statistically nonsignificant increase in 1-year mortality, and significantly in- ommendedininternationalguidelines(90 creased risk of the composite of mortality or readmission for acute myocardial infarc- minutes for PPCI and 30 minutes for tion or heart failure at 1 year. 1,2 fibrinolysis). In these registries, shorter JAMA. 2010;303(21):2148-2155 www.jama.com treatment delays are associated with bet- 5,8-10 teroutcomes andresultsofthe2treat- low-up, frequent exclusion of patients ments of important covariates).7,8,10,13,14 11,12 ments tend to be similar. transferred for PPCI, and residual con- An assessment of bias within a large pa- However, registry data also have limi- founding (lack of detailed measure- tient registry found that a substantial pro- tations, including selection bias, information bias (noninclusion of patients re- Author Affiliations: Agence d’e´ valuation des technolo- Que´ bec City, Canada (Drs Rodes-Cabau and Bogaty). ferred for PPCI but not receiving it), gies et des modes d’interventions en sante´ (Drs Lam- Corresponding Author: Laurie Lambert, PhD, Cardi- bert and Bogaty and Mr Brown), Jewish General Hos- ology Evaluation Unit, Agence d’e´ valuation des tech- incomplete and inappropriately short fol- pital, McGill University/Urgences-sante´ (Dr Segal), and nologies et des modes d’interventions en sante´ (Que- McGill University Health Center, McGill University (Dr bec Healthcare Assessment Agency), 2021 Union Ave, Brophy), Montreal, Que´ bec, Canada; and Institut uni- Montreal, QC H3A 2S9, Canada (laurie.lambert See also pp 2156 and 2188. versitaire de cardiologie et pneumologie de Que´ bec, @aetmis.gouv.qc.ca).

Downloaded From: https://jamanetwork.com/ on 10/01/2021 TIMELINESS OF THERAPY FOR ST-ELEVATION MI

portion of eligible patients were not en- and followed a standardized data collec- cluded more than once during the ob- rolled and that these patients received tion process to abstract all information servation period, only the first event was poorer-quality care and had 3-fold higher directly from medical charts and enter retained in the analysis. in-hospital mortality than enrolled pa- data onto a secure, centralized Web site tients.13,14 Finally, registries based on with automatic encryption. At the Variables and Outcomes voluntary participation of institutions completion of data entry, date-time val- Clinical factors included age, sex, earli- usually provide information from well- ues were independently validated by se- est in-hospital measure of systolic blood performing, motivated academic and nior medical librarians in a representa- pressure, heart rate at first in-hospital research-oriented centers but tend to tive 10% sample and concordance was ECG, and symptom duration. A TIMI in- miss an important proportion of pa- very high (median difference for door to dex was calculated for each patient: tients in the larger real-world setting. reperfusion therapy=0 min [interquar- (heart rateϫ[age/10]2)/systolic blood To address these issues, the Agence tile range {IQR}, 0-0]).15 pressure.16 The TIMI index was used in d’e´valuation des techniques et modes Study eligibility was assessed via an al- statistical analyses rather than the 3 fac- d’interventions en sante´, a government- gorithm in all patients with appropriate tors individually because it has been vali- supported but independent health care presenting symptoms, a final diagnosis dated as a powerful predictor of mortal- evaluation agency in Quebec, Canada, re- ofAMI(InternationalClassificationofDis- ity in STEMI patients and improved the cently undertook a systematic province- eases,NinthRevisioncode410),andahos- statistical power of the multivariate wide evaluation of STEMI care in Que- pital discharge date between October 1, model.16 Anterior STEMI and left bundle bec (population, 7.8 million). The 2006,andMarch31,2007.Patientswhose branch block were centrally identified objectives were to obtain a contempo- AMIoccurredaftertheirinitialemergency based on the admission ECG. rary portrait of reperfusion treatments department presentation were excluded. Treatment time was calculated from and their delays across Quebec and to de- ForeachpotentialSTEMIpatient,thefirst patient arrival at the first hospital (door) termine whether STEMI reperfusion electrocardiogram (ECG) taken after ar- to either start of fibrinolysis (needle) or treatment outside of the guideline- rival at the emergency department was first dilation (balloon). To ascertain pa- recommended delays is associated with sent to the coordinating center, where it tient comorbidities, the recorded princi- poorer outcomes than treatment within was interpreted by 2 cardiologists and an pal diagnosis as well as the 15 secondary recommended delays. emergency physician. Patients included diagnosesofallhospitaladmissionsinthe in the evaluation had to have 1 of the fol- 5 years preceding the date of admission METHODS lowing features: (1) received fibrinoly- for the index STEMI were identified in The evaluation covered a 6-month period sis in the 4 hours following triage at the Quebec’s administrative hospital dis- (October 1, 2006, to March 31, 2007) and firstemergencydepartment;(2)weresent charge database. We included diagnoses included all acute care Quebec hospitals to a catheterization laboratory in the 4 in the index STEMI admission only for (n=80) that treated at least 30 acute myo- hours following triage at the first emer- comorbidities considered to be chronic cardial infarctions (AMIs) in the preced- gency department and had mention of to avoid confusing an index admission ing year. These 80 hospitals treat more STEMI in the medical chart or, if not, evi- complication with a comorbidity. than 95% of all AMIs occurring in Que- denceofSTEMIonthefirstECGvalidated The primary end points were death at bec and are distributed across 15 of Que- at the coordinating center; or (3) did not 30 days, death at 1 year, and the com- bec’s18administrativehealthcareregions. receive fibrinolysis and were not sent to bined end point of death or hospital re- Some regions with small populations and acatheterizationlaboratoryinthe4hours admission for congestive heart failure or similar systems of care were combined in after triage at the first emergency depart- AMI at 1 year. Other end points in- our analyses (n=12 regions). ment but had mention of STEMI in the cluded PCI at 1 year (excluding the in- Ethics approval was obtained from medical chart and evidence of STEMI on dex admission) and coronary artery by- each of the 80 hospitals as well as by the first ECG validated at the coordinat- pass graft surgery at 1 year. All end points the provincial ethics board (Quebec ing center. were determined by linkage with pro- Commission for Access to Informa- In data analyses, all patients who sat- vincial medicoadministrative databases tion), which waived the need for pa- isfied criteria 1 and 2 as described were whose reliability has been shown.17,18 tient consent because this evaluation in- considered to have received reperfu- volved no intervention or patient sion treatment (intention to treat). How- Statistical Analysis contact and anonymity was ensured. ever, patients fulfilling criteria 2 who Mortality at 30 days and at 1 year was as- were sent for PPCI but who did not un- certained according to type of treatment Data Collection dergo PPCI were not evaluated in analy- (fibrinolysisorPPCI),treatmentdelayand and Patient Identification ses of timeliness of treatment. Patients timeliness (Յ30 minutes for fibrinolysis In each hospital, a certified medical rec- fulfilling criteria 3 were considered as not and Յ90 minutes for PPCI), clinical fac- ord librarian was designated, received an having received reperfusion treatment tors, and comorbidities. After adjustment individualized on-site training session, (no intention to treat). For patients infor age, no significant interaction was

Downloaded From: https://jamanetwork.com/ on 10/01/2021 TIMELINESS OF THERAPY FOR ST-ELEVATION MI

observed between any clinical risk factor randomeffect.21 Ofthe10variables,those In a third sensitivity analysis, we com- or comorbidity and type of treatment. referring to process of care were included pared mortality by type of treatment for Thus, the odds ratio (OR) and 95% con- as region-level variables (percentage of patients who were treated in study hos- fidence interval (CI) for the association patientsreceivingtimelyreperfusion,who pitals where there was an exclusive treat- between each variable and mortality were used an ambulance, or who were trans- ment practice (hospitals treating Ն95% calculated with the 2 treatment groups ferred for PPCI) while the remaining 7 of STEMI patients with either fibrinoly- combined. Statistical significance was riskfactorswereincludedattheindividual sis or PPCI). In these hospitals, the choice based on 2-sided testing with PϽ.05. patient level. Because of the inclusion of of reperfusion treatment is unlikely to To measure the association between patients without intention to treat in this have any association with patient char- timeliness of reperfusion (fibrinolysis and model,anadditionalregion-levelvariable acteristics. Statistical analyses were per- PPCI combined) and mortality, we used was added to the model representing the formed using SAS software25 and R soft- a multivariate logistic regression model percentageofpatientswhoreceivedreper- ware26 as well as the R add-ons lme427 that included timeliness of treatment, 3 fusion treatment. The model estimated and Matching.28 clinical factors (TIMI index, sex, and the region-level intercepts and their over- anterior MI), 4 comorbidities (conges- all variance. RESULTS tive heart failure, renal failure, periph- Due to the possibility of biased mul- Reperfusion Treatment Profile eral vascular disease, and cancer), and 2 tivariate effect estimates in nonexperi- We identified 6734 patients who pre- confounding factors (arrival by ambu- mental studies when the ratio of out- sented to an emergency department lance and transfer status) that have been comes to potential confounders is low, with both characteristic symptoms and shown to be associated with both time- a matched propensity score analysis was a final diagnosis of AMI, of whom 2361 liness of treatment and mortality.8,19,20 The used as a sensitivity analysis of the effect (35%) had STEMI. Five (0.2%) were ex- number of variables in the model was lim- of treatment timeliness.21,22 The propen- cluded because no link could be estab- ited to have a reasonable ratio of factors sity score was calculated by using 18 vari- lished with the other medicoadminis- to outcome events. The choice of variables (age, female sex, systolic blood pres- trative databases, leaving 2356 patients. ables was made a priori on the basis of sure Ͻ100 mm Hg, heart rate Ͼ100/ The majority of STEMI patients were consensual clinical judgment, estab- min, transfer status, mode of patient treated with PPCI (1440/2356 [61.1%]); lished associations with mortality,6,10,11 arrival, previous MI, previous chronic 16.6% (392/2356 patients) were treated and sufficient prevalence within the study heart failure, cardiac dysrhythmia, with fibrinolysis (tenecteplase in 99%) population (Ͼ5%). All 10 variables were peripheral vascular disease, chronic and 22.2% (524/2356 patients) neither included in each multivariate model. obstructive pulmonary disease, diabe- had fibrinolysis nor were sent for PPCI Given the number of STEMI patients tes, hypertension, cerebrovascular dis- within 4 hours of emergency depart- identified during the observation period, ease, cancer, renal failure, previous hos- ment arrival. Thus, of the patients treated we had 80% power to detect an OR of pital admission, and presence of Ն2 with acute reperfusion (n=1832), 78.6% 2.0 for the effect of timeliness on 30-day comorbidities) in a multivariate logistic (n=1440) were sent for PPCI and 21.4% mortality and an OR of 1.8 for 1-year regression for timeliness of treatment. (n=392) received fibrinolytic therapy. Of mortality. We had 80% power to detect Patients receiving timely treatment were the patients sent for PPCI, 34.7% an OR of 1.6 for the combined outcome. matched 1:1 with patients receiving late (n=499) were direct admissions to a To address concerns that low-risk pa- treatment if they had a propensity score PPCI center while 65.3% (n=941) were tients might be preferentially selected for within 0.1 of each other and in such a transferred from a non-PCI center. Of pa- reperfusion treatment, an analysis of 30- way that minimized total absolute dif- tients treated with fibrinolysis, 79.1% daymortalityattheregionallevelwasper- ferences in propensity scores.23 Match (n=310) were eventually sent for car- formed that included all STEMI patients balance was assessed by standardized and diac catheterization, the majority (53%) (with and without intention to treat). absolute differences on each covari- more than 12 hours later. Overall regional 30-day mortality was ate.24 Conditional logistic regression was Among patients referred for PPCI, plotted as a function of the regional pro- used to assess the OR and 95% CI of death 10.6% (n=153) did not undergo the proportion of patients treated within maxi- (at 30 days and at 1 year) and the com- cedure for various reasons, such as reso- mum recommended delays as well as a bined outcome for the matched pairs. lution of symptoms, a nonsignificantly function of the proportion of patients In another sensitivity analysis, we ex- narrowedculpritlesion,uncertaintyabout treatedwithPPCI.Anunadjustedinverse- amined timeliness and mortality by type theculpritlesion,inabilitytopasstheguide variance weighted least squares regres- of treatment for patients who, on the wire, or a decision to perform coronary sion was performed to ascertain the slope basis of the first ECG, were unani- artery bypass graft surgery instead. of each association. A multilevel logistic mously considered by the emergency regression analysis was conducted that physician and the 2 cardiologists at the Clinical Characteristics included the 10 variables as fixed effects coordinating center to have an un- Patient characteristics and comorbidities and the intercept term as a region-level equivocal STEMI. bytypeoftreatmentareshowninTABLE 1.

Downloaded From: https://jamanetwork.com/ on 10/01/2021 TIMELINESS OF THERAPY FOR ST-ELEVATION MI

The median age was 58 years in patients Outcomes patients with an admission ECG un- who received fibrinolysis and 60 years in Among patients who neither received fi- equivocal for STEMI were considered, PPCI patients (P=.01). The median TIMI brinolysis nor were sent for PPCI within mortality at 30 days was 4.5% for those indexwas2pointshigherinPPCIpatients 4 hours of triage, all-cause mortality was receiving fibrinolysis and 4.7% in those than in patients who received fibrinoly- 18% at 30 days and 29% at 1 year. TABLE 2 sent for PPCI (OR, 1.06; 95% CI, 0.52- sis(P=.003).Comparedwithpatientssent shows that all-cause mortality in the 2 2.12). At 1 year, mortality was 5.4% for forPPCI,morefibrinolytic-treatedpatients treatment groups was similar. At 30 days, those receiving fibrinolysis and 7.3% in presented to the emergency department 6.1% of patients who received fibrinoly- those sent for PPCI (OR, 1.51; 95% CI; within 3 hours of the start of symptoms sis and 5.6% of patients sent for PPCI had 0.81-2.81). In patients who actually un- whilefewerofthemarrivedbyambulance. died (OR, 0.91; 95% CI, 0.57-1.46). At derwent PPCI, mortality was 5.2% at 30 They were less likely than patients sent 1year,7.4%whoreceivedfibrinolysisand days and 7.9% at 1 year compared with for PPCI to have a history of arterial hy- 8.3% of patients sent for PPCI had died 9.2% and 11.7%, respectively, in those pertension or diabetes but more likely to (OR,1.13;95%CI,0.75-1.73).Whenonly who did not undergo the procedure. have a history of congestive heart failure. Table 1. Patient Characteristics According to Type of Treatmenta Timeliness of Reperfusion Fibrinolysis PPCI P Among patients who underwent PPCI, Characteristics (n = 392) (n = 1440) Value the median door-to-balloon time was 110 Demographic and clinical factors Age, median (IQR), y 58 (51-68) 60 (52-71) .01 minutes (IQR, 82-149 minutes); 32% had TIMI index, median (IQR)b 17.3 (12.7-24.9) 19.3 (13.7-27.8) .003 a delay within the maximum recom- Female 83 (21.2) 374 (26.0) .06 mended limit of 90 minutes. Among pa- Heart rate Ͼ100/min 39 (9.9) 158 (11.0) .63 tients who were directly admitted to a Systolic blood pressure Ͻ100 mm Hg 33 (8.4) 156 (10.8) .19 PPCI center, the median door-to- Anterior myocardial infarction 128 (32.6) 460 (31.9) .84 balloon time was 83 minutes (IQR, 61- Left bundle-branch block 8 (2.0) 31 (2.2) .95 117 minutes); 57% were treated within Symptom duration to triage Յ3 h 321 (81.9) 983 (68.3) Ͻ.001 90 minutes. For transferred patients, the Triage during work hours 211 (53.8) 727 (50.5) .26 median door-to-balloon time was 123 Transfer prior to reperfusion 0 941 (65.3) Ͻ.001 minutes (IQR, 98-160 minutes), and only Arrival by ambulance 223 (56.9) 905 (62.8) .04 19% were treated within 90 minutes. For Comorbidities Previous myocardial infarction 47 (12.0) 195 (13.5) .47 patients who received fibrinolysis, the Previous PCI 18 (4.6) 107 (7.4) .06 median delay was 33 minutes (IQR, Previous heart failure 18 (4.6) 32 (2.2) .02 20-52 minutes); 46% had a door-to- Peripheral vascular disease 23 (5.9) 93 (6.5) .76 needle time within the maximum rec- Chronic obstructive pulmonary disease 38 (9.7) 153 (10.6) .66 ommended delay of 30 minutes. Diabetes 49 (12.5) 249 (17.3) .03 We examined timeliness of treat- Renal disease 23 (5.9) 91 (6.3) .83 ment in patients who had an admis- Hypertension 137 (34.9) 598 (41.5) .02 sion ECG that was considered un- Cerebrovascular disease 11 (2.8) 50 (3.5) .62 equivocal for STEMI. These patients Cancer 16 (4.1) 57 (4.0) .97 constituted 52% (669/1287) of PPCI pa- Abbreviations: IQR, interquartile range; PCI, percutaneous coronary intervention; PPCI, primary percutaneous coronary intervention. tients and 62% (242/392) of fibrinoly- aData are expressed as No. (%) unless otherwise indicated. b 2 sis patients. Timeliness of treatment of TIMI index is calculated as (heart rateϫ[age/10] )/systolic blood pressure. PPCI-treated patients directly admitted to a PPCI center increased to 70% Table 2. Adverse Outcomes by Type of Treatment No. (%) (from 57%); for patients transferred for PPCI vs Fibrinolysis, PPCI, it increased to 25% (from 19%); Fibrinolysis PPCIa Unadjusted OR and for patients treated with fibrinoly- Outcomes (n = 392) (n = 1440) (95% CI) sis, timeliness of treatment increased Death at 30 d 24 (6.1) 81 (5.6) 0.91 (0.57-1.46) from 46% to 55%. Death at 1 y 29 (7.4) 120 (8.3) 1.13 (0.75-1.73) Readmission for heart failure at 1 y 11 (2.8) 50 (3.5) 1.22 (0.63-2.37) The 2 other factors most strongly as- Readmission for AMI at 1 y 14 (3.6) 36 (2.5) 0.69 (0.37-1.30) sociated with untimely reperfusion treat- Death or readmission for heart failure or AMI at 1 y 53 (13.5) 196 (13.6) 1.01 (0.73-1.40) ment in univariate analyses were trans- PCI at 1 y (post–index hospitalization) 25 (6.4) 78 (5.4) 0.84 (0.53-1.34) fer from a non-PCI center to a PPCI Coronary artery bypass graft surgery at 1 y 46 (11.7) 114 (7.9) 0.65 (0.45-0.93) center (OR, 4.63; 95% CI, 3.73-5.78) and Abbreviations: AMI, acute myocardial infarction; CI, confidence interval; OR, odds ratio; PPCI, primary percutaneous coro- arrival without use of an ambulance (OR, nary intervention. aIncludes all patients sent for PPCI whether or not PPCI was performed. 1.92; 95% CI, 1.56-2.38).

Downloaded From: https://jamanetwork.com/ on 10/01/2021 TIMELINESS OF THERAPY FOR ST-ELEVATION MI

About two-thirds of patients received Adverse Outcomes and Timeliness gestive heart failure or AMI at 1 year reperfusion therapy in hospitals where of Treatment occurred significantly more fre- choice of reperfusion treatment was ex- In an unadjusted analysis, for each quently when treatment was untimely clusive or nearly so (Ն95% either PPCI treatment type, mortality was signifi- among both patients who received fi- orfibrinolysis).Inexclusivetreatmenthos- cantly higher in patients treated out- brinolysis and those who underwent pitals, 30-day mortality in patients who side of recommended times than in PPCI (Table 3). Untimely fibrinolysis received fibrinolysis was 5.8% compared patients treated within recommended was associated with a worse prognosis with 5.0% in patients sent for PPCI (OR, times (TABLE 3). At 30 days, patients than timely PPCI, just as untimely PPCI 0.85; 95% CI, 0.45-1.63). At 1 year, mor- treated with PPCI outside of recom- was associated with a worse prognosis tality was 6.8% for patients who received mended times had higher mortality than timely fibrinolysis. When the 2 fibrinolysis and 7.3% for patients sent for (OR, 1.87; 95% CI, 1.02-3.41) as did treatment groups were combined, the PPCI (OR, 1.08; 95% CI, 0.60-1.96). patients treated outside of recom- risk of adverse events at 1 year was Incidence of the combined end point mended times with fibrinolysis (OR, higher (OR, 1.75; 95% CI, 1.26-2.41) (death or readmission for heart failure or 2.75; 95% CI, 1.07-7.08). The increase in patients who were not treated within AMI) at 1 year was 13.5% for fibrinoly- in mortality of untimely treatment was maximum recommended delays. sis patients and 13.6% for PPCI patients sustained at 1 year for both untimely When treatment delay beyond maxi- (OR, 1.01; 95% CI, 0.73-1.40) (Table 2). PPCI (OR, 1.71; 95% CI, 1.06-2.76) mum recommended guidelines was ex- Occurrence at 1 year of coronary artery and untimely fibrinolysis (OR, 2.41; amined in 30-minute intervals in pa- bypassgraftsurgerywaslowerinthePPCI 95% CI, 1.04-6.00). tients who underwent PPCI, there was groupthaninthefibrinolysisgroup(7.9% Within each treatment group, there an increasing risk of 30-day mortality vs 11.7%, respectively; OR, 0.65; 95% CI, was nonsignificantly less readmission with increasing treatment delay for both 0.45-0.93), while occurrence of PCI (af- for congestive heart failure with timely patients admitted directly to a PPCI center the index hospitalization) was simi- compared with untimely treatment ter (P=.003) and patients transferred for lar in the 2 groups (5.4% vs 6.4%, respec- (Table 3). Overall, the composite end PPCI (PϽ.001). tively; OR, 0.84; 95% CI, 0.53-1.34). point of death or readmission for con- Association Between Patient Characteristics and Mortality Table 3. Adverse Outcomes by Type and Timeliness of Treatment Univariate associations with 30-day Outcomes No. (%) of Patients OR (95% CI)a mortality are shown in TABLE 4. Clini- Fibrinolysis Յ30 min Ͼ30 min cal factors significantly associated with (n = 182) (n = 210) increased 30-day mortality were fe- Mortality at 30 d 6 (3.3) 18 (8.6) 2.75 (1.07-7.08) male sex and higher TIMI index. Each Mortality at 1 y 8 (4.4) 21 (10.0) 2.41 (1.04-5.60) comorbidity was associated with a sta- Readmission for AMI at 1 y 7 (3.3) 7 (3.8) 0.86 (0.29-2.51) tistically significant increase in risk of Readmission for CHF at 1 y 2 (1.1) 9 (4.3) 4.02 (0.86-18.90) death. Arrival by ambulance was asso- Mortality, CHF, or AMI at 1 y 16 (8.8) 37 (17.6) 2.22 (1.19-4.14) ciated with an increased risk of death, PPCIb Յ90 min Ͼ90 min while being transferred for PPCI was as- (n = 417) (n = 870) sociated with a lower risk of death that Mortality at 30 d 14 (3.4) 53 (6.1) 1.87 (1.02-3.41) was not statistically significant. Mortality at 1 y 23 (5.5) 79 (9.1) 1.71 (1.06-2.76) Readmission for AMI at 1 y 8 (1.9) 21 (2.4) 1.26 (0.56-2.88) Association Between Timeliness Readmission for CHF at 1 y 8 (1.9) 33 (3.8) 2.02 (0.92-4.40) of Treatment and Outcomes: Mortality, CHF, or AMI at 1 y 39 (9.4) 125 (14.4) 1.63 (1.11-2.38) Multivariate Analysis Combined Fibrinolysis and PPCIb Յ Recommended Ͼ Recommended After multivariate adjustment, esti- Delay Delay mates of the association between time- (n = 599) (n = 1080) liness of treatment and adverse events re- Mortality at 30 d 20 (3.3) 71 (6.6) 2.04 (1.22-3.38) 2.14 (1.21-3.93)c mained largely unchanged from Mortality at 1 y 31 (5.2) 100 (9.3) 1.87 (1.23-2.83) unadjusted estimates (Table 3). For mor- 1.61 (1.00-2.66)c tality in the 2 treatment groups com- Mortality, CHF, or AMI at 1 y 55 (9.2) 162 (15.0) 1.75 (1.26-2.41) bined, the adjusted relative risk associ- 1.57 (1.08-2.30)c ated with untimely treatment was 2.14 Abbreviations: AMI, acute myocardial infarction; CHF, congestive heart failure; CI, confidence interval; OR, odds ratio; PPCI, (95% CI, 1.21-3.93) at 30 days and 1.61 primary percutaneous coronary intervention. aOdds ratios are unadjusted unless otherwise indicated. (95% CI, 1.00-2.66) at 1 year. For the bPatients sent to the catheterization laboratory but who did not receive PPCI were excluded from these analyses. combined outcome of death or readmis- cAdjusted OR. sion for heart failure or AMI at 1 year,

Downloaded From: https://jamanetwork.com/ on 10/01/2021 TIMELINESS OF THERAPY FOR ST-ELEVATION MI

the adjusted relative risk for untimely COMMENT tion of patients treated with fibrinolysis treatment was 1.57 (95% CI, 1.08-2.30). In this evaluation, we sought to obtain and PPCI, particularly in patients trans- In the propensity analysis, of the 599 a complete contemporary portrait of ferred for PPCI; (3) risk of adverse events patients receiving timely treatment STEMI reperfusion treatment in Que- was similar in patients treated with PPCI (Table 3), a matched patient receiving bec and to ascertain the effect of un- and fibrinolysis but was higher in those late treatment was found for 504 patients timely treatment on patient outcomes. treated outside of the maximum recom- (84%; difference of Ͻ5% of standard de- The principal findings are that (1) PPCI mended delays, regardless of the reper- viation for all variables). The resultant is the predominant reperfusion strategy fusion strategy; and (4) timeliness of ORs from the matched propensity analy- in the province of Quebec, with most treatment with either reperfusion mode sis for 30-day mortality (OR, 2.00; 95% PPCI patients transferred from non- was a strong predictor of overall re- CI, 1.12-3.58), for 1-year mortality (OR, PCI hospitals; (2) delays to reperfusion gional mortality but no association was 1.59; 95% CI, 0.98-2.58), and for the treatment exceeded maximum recom- observed with choice of reperfusion treat- combined outcome (OR, 1.57; 95% CI, mended delays in a substantial propor- ment within health care regions. 1.07-2.29) were not substantially differ- ent than those yielded by the multivariate logistic regression models. Table 4. Univariate and Multivariate Results OR (95% CI) Timeliness of Reperfusion Treatment Mortality at 30 d Mortality, CHF, and Outcomes by Region Mortality at 1 y, or AMI at 1 y, a b c Across Quebec health care regions, Univariate Adjusted Adjusted Adjusted Female 2.58 (1.68-3.97) 1.82 (1.11-2.95) 1.77 (1.16-2.68) 1.47 (1.04-2.06) overall timeliness of treatment (PPCI TIMI index 1.08 (1.06-1.09) 1.07 (1.05-1.08) 1.06 (1.05-1.07) 1.05 (1.04-1.06) and fibrinolysis combined) ranged from Anterior AMI 1.47 (0.96-2.27) 1.51 (0.92-2.45) 1.46 (0.95-2.21) 1.82 (1.32-2.52) 19% to 76% and use of fibrinolysis and History of CHF 4.86 (1.99-10.76) 2.25 (0.81-5.71) 2.99 (1.27-6.70) 3.84 (1.84-7.90) PPCI each varied from 0% to 100% History of renal failure 5.11 (2.72-9.20) 1.49 (0.71-3.01) 2.27 (1.21-4.14) 2.27 (1.30-3.87) (FIGURE). In the unadjusted least History of peripheral 3.62 (2.00-6.58) 1.81 (0.84-3.68) 1.64 (0.83-3.10) 1.57 (0.88-2.72) squares regression analysis, each 10% vascular disease increase in patients treated within rec- History of cancer 3.25 (1.43-6.73) 1.34 (0.55-2.96) 2.62 (1.31-5.05) 1.82 (0.97-3.32) ommended time was associated with a Arrival by ambulance 3.53 (1.98-6.30) 2.03 (1.12-3.93) 2.05 (1.25-3.51) 1.21 (0.85-1.73) 1.09% decrease (95% CI, −1.76% to Transfer 0.71 (0.46-1.09) 0.54 (0.32-0.89) 0.84 (0.54-1.29) 0.83 (0.59-1.16) −0.42%) in region-level mortality at 30 Untimely treatment 2.04 (1.22-3.38) 2.14 (1.21-3.93) 1.61 (1.00-2.66) 1.57 (1.08-2.30) Abbreviations: AMI, acute myocardial infarction; CHF, congestive heart failure; CI, confidence interval; OR, odds ratio. days (Figure, A). However, there was aC statistic for multivariate model of 30-day mortality=0.86. bC statistic for multivariate model of 1-year mortality=0.85. no relationship between regional use of cC statistic for multivariate model of combined 1-year outcome=0.78. PPCI and region-level overall mortality at 30 days (slope, 0.0117; 95% CI, −0.046 to 0.070) (Figure, B). Figure. Regional 30-Day Mortality of All STEMI Patients by Intraregional Proportion of After adjustment for patient and Patients Treated Within Recommended Delays and Percentage of Patients Receiving Reperfusion With PPCI regional characteristics, the proportion of patients treated within recom- A Patients treated within recommended delay B Patients treated with PPCI mended time at the region level was significantly associated with the region’s 15 15 odds of death for all STEMI patients. Each 10% increase in regional timeliness was

associated with a region-level odds of 10 10 overall 30-day mortality of 0.80 (95% CI, 0.65-0.98). The other region-level variables in the multilevel model were not significant. Regional mortality in patients 5 5 who did not receive reperfusion treat-

ment was not higher in regions where y = –0.109× + 12.0; r = –0.75 y = 0.0117× + 6.92; r = 0.14 (95% CI, –0.92 to –0.31) (95% CI, –0.47 to 0.66)

mortality for patients receiving reperfu- Cumulative Incidence of Mortality at 30 d, % 0 Cumulative Incidence of Mortality at 30 d, % 0 sion treatment was low (slope, 0.54; 95% 0 20 40 60 80 100 0 20 40 60 80 100 CI, −1.39 to 2.46). This suggests that the Patients Treated Within Recommended Delay, % Patients Treated With PPCI, % relationship between timeliness of reper- Each circle represents a region or combined region. The size of the circles is proportional to the number of fusion and mortality was not due to selec- patients with ST-elevation myocardial infarction (STEMI) treated with reperfusion within the region. CI indi- cates confidence interval; PPCI, primary percutaneous coronary intervention. tive treatment of healthier patients.

Downloaded From: https://jamanetwork.com/ on 10/01/2021 TIMELINESS OF THERAPY FOR ST-ELEVATION MI

Our study, while consistent with reg- tion of STEMI care. Regardless of reper- reperfusion option and an untimely istry and clinical data associating longer fusion strategy, patients treated be- one, the former option appears associ- treatment delays with poorer out- yond maximum recommended delays ated with better patient outcomes. comes,8-10 is novel and robust in several had increased mortality. Occurrence of Consistent with results in treated pa- ways. Above all, it represents not a sam- congestive heart failure, an important tients only, we found an important and pling but more than 95% of all STEMI determinant of poor prognosis, was also statistically significant relation between patients within a large and complex sys- increased and is consistent with the no- regions’ timeliness of treatment and sur- tem of care and provides very recent tion that longer treatment times lead to vival in an analysis of all patients with information that transcends the rela- less myocardial salvage, making sub- STEMI. Regions that delivered reperfu- tive selectivity of randomized clinical sequent heart failure more likely. In sion treatment beyond maximum rec- trials and most registries. Importantly, contrast, longer treatment delay was not ommended delays had higher mortality patients transferred for PPCI, patients in associated with an increased risk of re- than regions that delivered reperfusion whom there was no initial intention to admission for AMI, which may not be treatment within recommended delays treat with reperfusion, and patients who unexpected since there is no plausible while mode of reperfusion within the re- were sent for reperfusion treatment but pathophysiological link between longer gion was not associated with mortality. who did not receive it for various rea- treatment delay and risk of new MI. This regional analysis of all STEMI pa- sons were all included. Patients with The association between longer delay tients suggests that the observed ben- STEMI were systematically identified on to reperfusion treatment and poorer out- efits of timely treatment are not due to the basis of an algorithm and central come may be confounded because a selection bias of treating healthier pa- interpretation of the admission ECG. patients who are “sicker” may be treated tients or treating them more quickly. Linkage with medicoadministrative data- later rather than earlier. For example, These results may reassure clinicians and bases permitted assessment of comor- hemodynamically unstable patients could health care policy planners in regions bidities and ensured completeness of fol- require more time to be stabilized before where PPCI is not an option that timely low-up beyond the short term not only being able to undergo reperfusion treat- fibrinolysis is a very acceptable stan- of mortality but also of the important end ment. Also, patients unlikely to survive dard of STEMI care. points of readmission for AMI or heart a long delay to reperfusion may be less Although retrospective collection of failure and revascularization proce- likely to be transferred for PPCI and data has limitations, it has the impor- dures. Sensitivity analyses, notably might receive no reperfusion treatment. tant advantage of allowing a more com- restricted to patients with an ECG However, after multivariate- plete and systematic process for STEMI unequivocal for STEMI and centers per- adjusted analyses that included adjust- patient identification, thus minimiz- forming exclusive reperfusion treatment for clinical risk factors (TIMI in- ing some biases.13,14 The accuracy of ments, as well as propensity analysis and dex, female sex) and comorbidities data entry concerning the process of region-level analysis, strengthen confi- (history of heart failure, peripheral vas- STEMI care was also found to be high15 dence in the observed associations. cular disease, renal failure, or cancer) but this observational study has limi- Moreover, since this study was per- as well as factors known to be associ- tations. Nonrecorded data might have formed in a single-payer, universal access ated with both timeliness and mortal- led to residual confounding that could system,observedassociationsareunlikely ity (presentation by ambulance and have influenced our findings. How- to be confounded by patient differences transfer status),8,19,20 an increase in ad- ever, adjustment for available con- in insurance and socioeconomic status. verse outcomes in patients with longer founding variables, including impor- Thus, we believe this evaluation repre- delays to reperfusion was observed for tant clinical factors, did not materially sents a needed contribution to the evi- mortality in the short term and for a change observed results. Although dence base for deriving clinical practice combined measure of morbidity and medicoadministrative databases en- guidelines29 and an important advance in mortality in the long term. Moreover, sured completeness of follow-up for the knowledge of the outcomes associated a propensity analysis that matched an primary end points, they did not con- with contemporary processes of STEMI array of variables in patients with un- tain appropriate information to assess care. This “real-world” information is rel- timely treatment to patients with timely hemorrhagic events. evant both clinically and from a perspec- treatment resulted in similar observa- Among patients in Quebec with tive of evidence-based health care policy tions of increased risk. Thus, this prov- STEMI, reperfusion delivered outside and planning,14 pointing to the lifesav- ince-wide evaluation provides up-to- of guideline-recommended treatment ing potential for approaches that focus date, real-world evidence for the delays was associated with signifi- on offering the most timely reperfusion emphasis expressed in international cantly increased 30-day mortality, a sta- treatment to patients with STEMI. guidelines on the importance of con- tistically nonsignificant increase in Time, rather than mode of reperfu- sidering timeliness of treatment when 1-year mortality, and significantly in- sion, emerges as a critical determinant choosing the mode of reperfusion.1,2 creased risk of the composite out- of outcome in this systematic evalua- Faced with a choice between a timely come of mortality or readmission for

Downloaded From: https://jamanetwork.com/ on 10/01/2021 TIMELINESS OF THERAPY FOR ST-ELEVATION MI

AMI or heart failure at 1 year. The study REFERENCES farction (Vienna STEMI Registry). Circulation. 2006; 113(20):2398-2405. findings suggest that time to reperfu- 1. Van de Werf F, Bax J, Betriu A, et al; ESC Com- 13. Ferreira-Gonza´ lez I, Marsal JR, Mitjavila F, et al. sion rather than treatment strategy may mittee for Practice Guidelines. Management of acute Patient registries of acute coronary syndrome: assess- myocardial infarction in patients presenting with per- ing or biasing the clinical real world data? Circ Car- be more important in terms of out- sistent ST-segment elevation: the Task Force on the diovasc Qual Outcomes. 2009;2(6):540-547. comes and can help inform clinical de- Management of ST-Segment Elevation Acute Myo- 14. Krumholz HM. Registries and selection bias: the cardial Infarction of the European Society of Cardiology. need for accountability. Circ Cardiovasc Qual cision making to optimize care for pa- Eur Heart J. 2008;29(23):2909-2945. Outcomes. 2009;2(6):517-518. tients with AMI presenting to hospitals 2. Kushner FG, Hand M, Smith SC Jr, et al; Ameri- 15. Lambert LJ, Carroll-Bilodeau C, Giguère M, et al. in Quebec. can College of Cardiology Foundation/American Heart Reliability of process measures of care of patients pre- Association Task Force on Practice Guidelines. 2009 senting with ST-elevation myocardial infarction (STEMI) Author Contributions: Drs Lambert and Bogaty and Focused updates: ACC/AHA guidelines for the man- abstracted by hospital health record librarians: re- Mr Brown had full access to all of the data in the study agement of patients with ST-elevation myocardial in- sults of a field evaluation in Quebec, Canada [ab- and take responsibility for the integrity of the data and farction (updating the 2004 guideline and 2007 fo- stract 175]. Circ Cardiovasc Qual Outcomes. 2009; the accuracy of the data analysis. cused update) and ACC/AHA/SCAI guidelines on 2(3):e40. Study conception and design: Lambert, Bogaty. percutaneous coronary intervention (updating the 16. Wiviott SD, Morrow DA, Frederick PD, et al. Per- Acquisition of data: Lambert, Bogaty, Segal. 2005 guideline and 2007 focused update): a report formance of the Thrombolysis in Myocardial Infarc- of the American College of Cardiology Foundation/ Analysis and interpretation of data : Brown, Lambert, tion risk index in the National Registry of Myocardial American Heart Association Task Force on Practice Bogaty, Segal, Brophy, Rodes-Cabau. Infarction-3 and -4: a simple index that predicts mor- Guidelines. Circulation. 2009;120(22):2271-2306. Drafting of the manuscript: Lambert, Bogaty, Brown. tality in ST-segment elevation myocardial infarction. 3. Huynh T, Perron S, O’Loughlin J, et al. Compari- Critical revision of the manuscript for important in- J Am Coll Cardiol. 2004;44(4):783-789. son of primary percutaneous coronary intervention and tellectual content: Segal, Brophy, Rodes-Cabau. 17. Levy AR, Tamblyn RM, Fitchett D, McLeod PJ, fibrinolytic therapy in ST-segment-elevation myocar- Hanley JA. Coding accuracy of hospital discharge data Statistical analysis: Brown, Lambert, Bogaty, Brophy, dial infarction: Bayesian hierarchical meta-analyses of Segal, Rodes-Cabau. for elderly survivors of myocardial infarction. Can J randomized controlled trials and observational studies. Cardiol. 1999;15(11):1277-1282. Obtained funding: Lambert, Bogaty. Circulation . 2009;119(24):3101-3109. 18. Thibault N. Suivi et cycle des naissances et des Administrative, technical, or material support: Lambert, 4. Steg PG, Lopez-Sendon J, Lopez de Sa E, et al; de´cès: qu’en est-il au Que´bec? Que´ bec, QC: Institut Bogaty, Brown, Segal, Brophy, Rodes-Cabau. GRACE Investigators. External validity of clinical trials de la Statistique du Que´ bec; 2007. http://www.bdso Study supervision: Lambert, Bogaty. in acute myocardial infarction. Arch Intern Med. 2007; .gouv.qc.ca/docs-ken/multimedia/PB01600FR Financial Disclosures: Dr Rodes-Cabau reports that 167(1):68-73. _Ecostat2007M03F02.pdf. Accessed April 16, 2010. he is recipient of a research grant from Boston Scien- 5. Nallamothu BK, Wang Y, Magid DJ, et al; Na- 19. So DY, Ha AC, Turek MA, et al. Comparison of tific and lecture fees from Abbott, Cordis, and Bos- tional Registry of Myocardial Infarction Investigators. mortality patterns in patients with ST-elevation myo- ton Scientific. No other disclosures were reported. Relation between hospital specialization with pri- cardial infarction arriving by emergency medical ser- Funding/Support: This evaluation was funded by the mary percutaneous coronary intervention and clini- vices vs self-transport (from the prospective Ottawa Ministry of Health of the province of Quebec. cal outcomes in ST-segment elevation myocardial in- Am J Cardiol Role of the Sponsor: The funder had no role in the farction: National Registry of Myocardial Infarction-4 Hospital STEMI Registry). . 2006; design and conduct of the study, analysis and inter- analysis. Circulation. 2006;113(2):222-229. 97(4):458-461. pretation of the data, or preparation, review, or ap- 6. Mehta RH, Bufalino VJ, Pan W, et al; American Heart 20. Westfall JM, Kiefe CI, Weissman NW, et al. Does proval of the manuscript. Association Get With the Guidelines Investigators. interhospital transfer improve outcome of acute myo- Additional Contributions: We dedicate this article to Achieving rapid reperfusion with primary percutane- cardial infarction? a propensity score analysis from the the memory of Konrad Jamrozik, MD, PhD. We grate- ous coronary intervention remains a challenge: in- Cardiovascular Cooperative Project. BMC Cardio- fully acknowledge the important contributions of Ce´- sights from American Heart Association’s Get With the vasc Disord. 2008;8:22. line Carroll, BSc, and Maude Giguère, AMA, Agence Guidelines program. Am Heart J. 2008;155(6): 21. Stukel TA, Fisher ES, Wennberg DE, Alter DA, d’e´ valuation des technologies et des modes 1059-1067. Gottlieb DJ, Vermeulen MJ. Analysis of observational studies in the presence of treatment selection d’interventions en sante´ (study conception and de- 7. Eagle KA, Nallamothu BK, Mehta RH, et al; Global bias: effects of invasive cardiac management on AMI sign, acquisition of data, administrative, technical, and Registry of Acute Coronary Events (GRACE) survival using propensity score and instrumental vari- material support, and training of 80 designated medi- Investigators. Trends in acute reperfusion therapy for able methods. JAMA. 2007;297(3):278-285. cal record librarians). We also gratefully acknowl- ST-segment elevation myocardial infarction from 1999 22. Cepeda MS, Boston R, Farrar JT, Strom BL. Com- edge other important contributions of the following to 2006: we are getting better but we have got a long parison of logistic regression vs propensity score when persons: Luce Boyer, RN, Centre de recherche de way to go. Eur Heart J. 2008;29(5):609-617. the number of events is low and there are multiple l’institut universitaire de cardiologie et de pneumolo- 8. Gibson CM, Pride YB, Frederick PD, et al. Trends confounders. Am J Epidemiol. 2003;158(3):280- gie de Que´ bec (conception and design); Peter Faris, in reperfusion strategies, door-to-needle and door- 287. PhD, Alberta Bone and Joint Institute (statistical ad- to-balloon times, and in-hospital mortality among patients with ST-segment elevation myocardial infarc- 23. Rosenbaum PR. Optimal matching for observa- vice on propensity analyses and statistical power cal- tional studies. J Am Stat Assoc. 1989;84(408): culation); Jean E. Morin, MD, McGill University Health tion enrolled in the National Registry of Myocardial Infarction from 1990 to 2006. Am Heart J. 2008; 1024-1032. Centre (conception and design); Jean-Marie Mout- 24. Austin PC. Assessing balance in measured base- quin, MD, Agence d’e´ valuation des technologies et 156(6):1035-1044. 9. Nallamothu B, Fox KA, Kennelly BM, et al; GRACE line covariates when using many-to-one matching on des modes d’interventions en sante´ (critical revision Investigators. Relationship of treatment delays and the propensity-score. Pharmacoepidemiol Drug Saf. of manuscript); James Nasmith, MD, St Paul’s Hos- mortality in patients undergoing fibrinolysis and pri- 2008;17(12):1218-1225. pital (analysis of data); Serge Simard, MSc, Centre de mary percutaneous coronary intervention: the Global 25. SAS Institute Inc. SAS Version 9.1.3. Cary, NC: recherche de l’institut universitaire de cardiologie et Registry of Acute Coronary Events. Heart. 2007; SAS Institute Inc; 2005. de pneumologie de Que´ bec (statistical advice and 93(12):1552-1555. 26. R Development Core Team. R: a language and analysis); Benedict Stuber-Gaumond, BSc, Centre de 10. Rathore SS, Curtis JP, Chen J, et al; National Car- environment for statistical computing: reference in- recherche de l’institut universitaire de cardiologie et diovascular Data Registry. Association of door-to- dex, version 2.10.0. Vienna, Austria: R Foundation for de pneumologie de Que´ bec (creation of Web site and balloon time and mortality in patients admitted to hos- Statistical Computing; 2009. http://www.r-project corresponding database); Re´ my The´ riault, PhD, Cen- pital with ST elevation myocardial infarction: national .org/. Accessed April 16, 2010. tre de recherche de l’institut universitaire de cardiolo- cohort study. BMJ. 2009;338:b1807. 27. Bates D, Maechler M. lme4: LINEAR mixed ef- gie et de pneumologie de Que´ bec (creation of Web 11. Danchin N, Coste P, Ferrieres J, et al; FAST-MI fects models using S4 classes (R package version site and corresponding database); and Jack V. Tu, MD, Investigators. Comparison of thrombolysis followed 0.999375-32). 2009. http://cran.r-project.org/web PhD, Institute for Clinical Evaluative Sciences (critical by broad use of percutaneous coronary intervention /packages/lme4/index.html. Accessed April 16, 2010. revision of manuscript). Ms Boyer, Drs Morin, Mout- with primary percutaneous coronary intervention for 28. Sekhon JS. Matching: multivariate and propen- quin, Nasmith, and The´ riault, and Messrs Simard and ST-segment-elevation acute myocardial infarction: data sity score matching with balance optimization (R pack- Stuber-Gaumond were compensated for their contri- from the French Registry on Acute ST-Elevation Myo- age version 4.7-6). Berkeley: University of California; butions to the project. We also thank the designated cardial Infarction (FAST-MI). Circulation. 2008; 2009. http://cran.r-project.org/web/packages medical record librarians in the 80 participating hos- 118(3):268-276. /Matching/index.html. Accessed April 16, 2010. pitals across Quebec for acquisition of data and the 12. Kalla K, Christ G, Karnik R, et al; Vienna STEMI 29. Tricoci P, Allen JM, Kramer JM, Califf RM, Smith Executive Committee and Scientific Committee of the Registry Group. Implementation of guidelines im- SC Jr. Scientific evidence underlying the ACC/AHA Quebec Tertiary Cardiology Network (our clinical part- proves the standard of care: the Viennese Registry on clinical practice guidelines. JAMA. 2009;301(8): ners in this endeavor). Reperfusion Strategies in ST-Elevation Myocardial In- 831-841.

Downloaded From: https://jamanetwork.com/ on 10/01/2021