Prognostic Implications of Procedural vs Spontaneous MI
Prognostic Implications of Procedural vs Spontaneous MI
The population for our study was drawn from the EVENT registry. As previously described, EVENT enrolled unselected patients undergoing percutaneous coronary intervention (PCI) with an approved intracoronary stent at more than 50 US centers between July 2004 and June 2007. Enrollment of patients was performed consecutively during specified recruitment "Waves" (eg, on predetermined days of the week) to minimize selection bias. The study protocol was approved by the institutional review board at each participating institution, and written informed consent was obtained from all patients before participation.
For the purpose of this study, patients enrolled in the EVENT registry with or without acute coronary syndrome (ACS; ST-segment elevation MI [STEMI]/non–ST-segment elevation MI [NSTEMI] or unstable angina) were chosen. Among the 10,148 patients enrolled in the EVENT registry, we excluded 1,839 patients missing either creatinine kinase–muscle brain fraction (CK-MB) or troponin values after the procedure, 802 patients with missing baseline biomarkers, and 127 patients with elevated biomarkers at baseline in the absence of an ACS, leaving a cohort of 7,380 patients. Patients were stratified into 3 groups (Figure 1): (1) no-MI group, patients who presented without ACS and with no elevation in baseline biomarkers (cardiac troponin [cTn I or T] or CK-MB <1× upper limit of normal [ULN]) and with peak post-PCI biomarkers ≤3× ULN; (2) spontaneous MI group, patients who presented with MI <7 days as the indication for the index PCI procedure; and (3) procedural MI group, patients who presented without MI but who experienced a procedural MI after PCI. Procedural MI was defined using the 2007 Universal Definition of MI (type 4a) as postprocedure peak cTn or CK-MB >3× ULN obtained 6 to 24 hours after PCI. Patients with spontaneous MI who had procedural MI were considered in the spontaneous MI group.
(Enlarge Image)
Figure 1.
Study design.
Data regarding baseline demographics, risk factors, cardiovascular history, angiographic characteristics, PCI procedural details (including angiographic complications), and clinical outcomes were collected prospectively on standardized case report forms by trained study coordinators and submitted to the data coordinating center (Harvard Clinical Research Institute, Boston, MA). Cardiac troponin (I or T), creatinine phosphokinase, and CK-MB levels were assessed at baseline (within 1 hour before the procedure) and every 8 hours, for a minimum of 2 samples, after the procedure and assayed using the clinical laboratory and reference values for each site. These values were normalized to the individual clinical center's ULN and are reported as a ratio. If an MI was suspected clinically at a later time point, additional biomarkers were obtained as clinically indicated. Patients were contacted by telephone at 6 and 12 months after the index PCI. All clinical outcomes were adjudicated by 2 cardiologists blinded to baseline variables.
The primary outcome for this analysis was mortality at 12 months. The secondary outcome was cardiovascular mortality at 12 months.
Continuous variables are reported as mean value ± SD and were compared across groups using Student t test or analysis of variance (for normally distributed variable) or the Wilcoxon rank sum test (for other variables). Categorical variables are reported as counts and proportions and were compared using the χ test.
Time-to-event variables were described using Kaplan-Meier estimates and compared using the log-rank statistic. To adjust for baseline differences among the 3 patient groups, we used a multiple propensity score approach. A propensity score is the conditional probability of having a particular exposure given a set of measured baseline covariates and is an established method for reducing confounding in observational studies. Although propensity score matching or stratification has been used to assemble patient cohorts that are similar in terms of baseline covariates for 2 treatment comparisons, this approach is problematic when there are more than 2 groups to be compared. In such cases, a multiple propensity score approach has been used as a solution to the "dimensionality problem." A multiple propensity score is defined as the conditional probability of falling into a particular group given a set of observed baseline covariates.
For our study, a multiple propensity score was estimated using a nonparsimonious multinomial logistic regression model with MI status (none, procedural, or spontaneous) as the dependent variable and the baseline covariates outlined in Table I as independent variables. The baseline covariates were then adjusted for the propensity scores. A Cox proportional hazards model was used to estimate the effect of comparator groups on the primary and secondary outcomes after adjusting for the propensity scores (p1, p2, p3) and their products (ie, product of any 2 of the 3 probabilities). A P value <.05 was considered statistically significant. In a sensitivity analysis, the primary results were rerun using a traditional proportional hazards regression model adjusted to baseline covariates. Further sensitivity analysis was performed after excluding patients who presented with a STEMI. All analyses were performed using IBM SPSS for Windows, Version 20.0.0 (IBM Corp, Chicago, IL).
Methods
Patient Population
The population for our study was drawn from the EVENT registry. As previously described, EVENT enrolled unselected patients undergoing percutaneous coronary intervention (PCI) with an approved intracoronary stent at more than 50 US centers between July 2004 and June 2007. Enrollment of patients was performed consecutively during specified recruitment "Waves" (eg, on predetermined days of the week) to minimize selection bias. The study protocol was approved by the institutional review board at each participating institution, and written informed consent was obtained from all patients before participation.
For the purpose of this study, patients enrolled in the EVENT registry with or without acute coronary syndrome (ACS; ST-segment elevation MI [STEMI]/non–ST-segment elevation MI [NSTEMI] or unstable angina) were chosen. Among the 10,148 patients enrolled in the EVENT registry, we excluded 1,839 patients missing either creatinine kinase–muscle brain fraction (CK-MB) or troponin values after the procedure, 802 patients with missing baseline biomarkers, and 127 patients with elevated biomarkers at baseline in the absence of an ACS, leaving a cohort of 7,380 patients. Patients were stratified into 3 groups (Figure 1): (1) no-MI group, patients who presented without ACS and with no elevation in baseline biomarkers (cardiac troponin [cTn I or T] or CK-MB <1× upper limit of normal [ULN]) and with peak post-PCI biomarkers ≤3× ULN; (2) spontaneous MI group, patients who presented with MI <7 days as the indication for the index PCI procedure; and (3) procedural MI group, patients who presented without MI but who experienced a procedural MI after PCI. Procedural MI was defined using the 2007 Universal Definition of MI (type 4a) as postprocedure peak cTn or CK-MB >3× ULN obtained 6 to 24 hours after PCI. Patients with spontaneous MI who had procedural MI were considered in the spontaneous MI group.
(Enlarge Image)
Figure 1.
Study design.
Data Collection and Follow-up
Data regarding baseline demographics, risk factors, cardiovascular history, angiographic characteristics, PCI procedural details (including angiographic complications), and clinical outcomes were collected prospectively on standardized case report forms by trained study coordinators and submitted to the data coordinating center (Harvard Clinical Research Institute, Boston, MA). Cardiac troponin (I or T), creatinine phosphokinase, and CK-MB levels were assessed at baseline (within 1 hour before the procedure) and every 8 hours, for a minimum of 2 samples, after the procedure and assayed using the clinical laboratory and reference values for each site. These values were normalized to the individual clinical center's ULN and are reported as a ratio. If an MI was suspected clinically at a later time point, additional biomarkers were obtained as clinically indicated. Patients were contacted by telephone at 6 and 12 months after the index PCI. All clinical outcomes were adjudicated by 2 cardiologists blinded to baseline variables.
Outcomes
The primary outcome for this analysis was mortality at 12 months. The secondary outcome was cardiovascular mortality at 12 months.
Statistical Analysis
Continuous variables are reported as mean value ± SD and were compared across groups using Student t test or analysis of variance (for normally distributed variable) or the Wilcoxon rank sum test (for other variables). Categorical variables are reported as counts and proportions and were compared using the χ test.
Time-to-event variables were described using Kaplan-Meier estimates and compared using the log-rank statistic. To adjust for baseline differences among the 3 patient groups, we used a multiple propensity score approach. A propensity score is the conditional probability of having a particular exposure given a set of measured baseline covariates and is an established method for reducing confounding in observational studies. Although propensity score matching or stratification has been used to assemble patient cohorts that are similar in terms of baseline covariates for 2 treatment comparisons, this approach is problematic when there are more than 2 groups to be compared. In such cases, a multiple propensity score approach has been used as a solution to the "dimensionality problem." A multiple propensity score is defined as the conditional probability of falling into a particular group given a set of observed baseline covariates.
For our study, a multiple propensity score was estimated using a nonparsimonious multinomial logistic regression model with MI status (none, procedural, or spontaneous) as the dependent variable and the baseline covariates outlined in Table I as independent variables. The baseline covariates were then adjusted for the propensity scores. A Cox proportional hazards model was used to estimate the effect of comparator groups on the primary and secondary outcomes after adjusting for the propensity scores (p1, p2, p3) and their products (ie, product of any 2 of the 3 probabilities). A P value <.05 was considered statistically significant. In a sensitivity analysis, the primary results were rerun using a traditional proportional hazards regression model adjusted to baseline covariates. Further sensitivity analysis was performed after excluding patients who presented with a STEMI. All analyses were performed using IBM SPSS for Windows, Version 20.0.0 (IBM Corp, Chicago, IL).
Source...