Author + information
- Received May 12, 2016
- Accepted June 2, 2016
- Published online September 12, 2016.
- Neeraj Shah, MD, MPHa,∗ (, )
- Rahul Chaudhary, MDb,
- Kathan Mehta, MD, MPHc,
- Vratika Agarwal, MDd,
- Jalaj Garg, MDa,
- Ronald Freudenberger, MDa,
- Larry Jacobs, MDa,
- David Cox, MDa,
- Karl B. Kern, MDe and
- Nainesh Patel, MDa
- aLehigh Valley Health Network, Allentown, Pennsylvania
- bJohns Hopkins University/Sinai Hospital of Baltimore, Baltimore, Maryland
- cUniversity of Pittsburgh Medical Center at Shadyside, Pittsburgh, Pennsylvania
- dStaten Island University Hospital, Staten Island, New York
- eUniversity of Arizona College of Medicine, Tucson, Arizona
- ↵∗Reprint Requests and Correspondence:
Dr. Neeraj Shah, Lehigh Valley Health Network, Department of Cardiology, 1250 South Cedar Crest Boulevard, Allentown, Pennsylvania 18103.
Objectives This study sought to determine whether “real-world” data supported the hypothesis that therapeutic hypothermia (TH) led to increased rates of stent thrombosis.
Background TH, which is often instituted after cardiac arrest (CA) to improve neurologic outcomes, alters pharmacokinetics of antiplatelet medications, leading to a theoretical risk of stent thrombosis after percutaneous coronary intervention (PCI).
Methods CA patients with acute myocardial infarction undergoing PCI were identified from the Nationwide Inpatient Sample from 2006 to 2011, with a defined primary outcome of stent thrombosis. The incidence of stent thrombosis in patients undergoing TH versus those not undergoing TH was compared using both logistic regression and propensity score matching.
Results In this dataset, 49,109 CA patients underwent PCI for acute myocardial infarction from 2006 to 2011, of whom 1,193 (2.4%) underwent TH. The incidence of stent thrombosis in the TH group was 3.9% (43 of 1,193), compared to 4.7% (2,271 of 47,916) in the no TH group (p = 0.61). Logistic regression showed that TH was not a significant predictor of stent thrombosis with an adjusted odds ratio of 0.71 (95% confidence interval: 0.28 to 1.76; p = 0.46). Propensity matching was performed to adjust for baseline differences between the TH and no TH groups, matching 1,155 patients in the TH group with 3,399 patients in the no TH group. No difference was observed in the incidence of stent thrombosis in the TH and the no TH groups after propensity matching (3.5% vs. 6.1%; p = 0.17).
Conclusions TH does not increase the incidence of stent thrombosis after primary PCI in patients with acute myocardial infarction presenting as CA.
There are approximately 347,000 to 362,000 cases of out-of-hospital cardiac arrest (OHCA) in the United States every year (1). Of these, few have return of spontaneous circulation and make it to the hospital. Of the patients who survive the initial event and are admitted to the hospital, 70% die before discharge and only two-thirds of the remainder are discharged with a good neurologic status (1,2). In 2014, the survival to hospital discharge of all nontraumatic emergency medical services–treated adult OHCA patients with any first recorded rhythm was only 12% (1). Approximately 20% to 30% of patients presenting with OHCA have coronary artery occlusion or an unstable coronary lesion, even in the absence of ST-segment elevation (3,4). Therapeutic hypothermia (TH) and emergent coronary angiography with primary percutaneous coronary intervention (PCI) in the appropriate patient population are one of the few interventions proven to improve outcomes in the setting of OHCA occurring due to a coronary event (3,5–7). The American Heart Association and the European Resuscitation Council recommend TH for all patients who remain unresponsive after return of spontaneous circulation, regardless of initial rhythm after cardiac arrest (CA) (8). TH causes reduction of cerebral glucose and oxygen metabolism, possibly providing neurologic protection, although the exact mechanism remains unknown (9).
The effects of hypothermia on blood coagulability have been debated and, although conventional wisdom points toward an increased risk of bleeding with hypothermia, recent studies have reflected increased thrombogenicity and platelet activity with hypothermia (10). As discussed, patients with OHCA often require primary PCI if a coronary event was responsible for the CA. Alterations in platelet reactivity and pharmacokinetics of antiplatelet agents with TH can theoretically pre-dispose the patients to a higher risk of stent thrombosis after PCI. Any beneficial effect of TH may thus be potentially offset by a higher incidence of stent thrombosis. There are few studies in the literature about stent thrombosis after PCI in CA patients, and there is no consistent relationship between TH and stent thrombosis in these studies (11–15). Given the paucity of data and lack of a consensus opinion of whether TH is associated with an increased incidence of stent thrombosis, we designed a study using the Nationwide Inpatient Sample (NIS), the largest all-payer in-patient database in the United States, to evaluate this association.
The NIS contains all discharge data from an approximate 20% stratified sample of U.S. hospitals. It is a part of the Healthcare Quality and Utilization Project, sponsored by the Agency for Healthcare Research and Quality. Data from the NIS have been used to identify, track, and analyze national trends in health care use, patterns of major procedures, access, disparity of care, trends in hospitalizations, charges, quality, and outcomes (16,17). Each individual hospitalization is de-identified and maintained in NIS as a unique entry with 1 primary discharge diagnosis and ≤25 secondary diagnoses during that hospitalization. It also incorporates 1 primary procedure code and ≤15 secondary procedure codes. Each entry also carries information on demographic details, insurance status, comorbidities, hospitalization outcome, and length of stay (LOS).
Validation and quality control
Annual data quality assessments of the NIS are performed, which guarantee the internal validity of the database. Furthermore, estimates from the NIS are compared with American Hospital Association Annual Survey Database, the National Hospital Discharge Survey from the National Center for Health Statistics, and the MedPAR inpatient database from Centers for Medicare and Medicaid Services (18). These reports are published in the NIS website and show that NIS resembled typical hospitals in American Hospital Association universe in most characteristics, all NIS and National Hospital Discharge Survey estimates agreed in overall and regional comparisons, and NIS Medicare measures were consistent with MedPar statistics. These reports strengthen the external validity of NIS database. Detailed reports regarding the data comparisons and data quality of NIS are available at the following website: http://www.hcup-us.ahrq.gov/db/nation/nis/nisrelatedreports.jsp. To ensure that the imported data were accurate, we verified the imported data against the Healthcare Cost and Utilization Project standard, available from the Healthcare Cost and Utilization Project website (http://www.hcupnet.ahrq.gov).
The NIS was queried from 2006 to 2011 using the International Classification of Diseases-9th Revision (ICD-9) diagnoses codes of 427.5 and V12.53 for CA in any diagnosis field. Acute myocardial infarction (AMI), including both ST-segment elevation myocardial infarction (STEMI) and non-STEMI, was identified using ICD-9 diagnoses codes 410.0 to 410.9 and coronary stent placement was identified using ICD-9 procedure codes 36.06 and 36.07. All CA patients with AMI undergoing coronary stent placement were included in our study population. TH was identified using ICD-9 procedure code 99.81, and the patients were divided into TH and no TH groups.
All patients >18 years of age after a resuscitated CA from AMI who underwent coronary stent placement were included.
Patients with intracranial bleeding (ICD-9 codes 430, 431, 432.0, 432.1, 432.9 in any diagnosis field) and severe sepsis (ICD-9 code 995.92 in any diagnosis field) were excluded, because these conditions are contraindications to the institution of TH. Patients with a primary diagnosis of stent thrombosis (ICD-9 code 996.72) were also excluded.
Stent thrombosis occurring during the same hospitalization was the primary outcome. It was identified using ICD-9 code 996.72 in any of the secondary diagnosis fields.
Comorbidities and PCI-related variables
Comorbidities such as coronary artery disease, hypertension, diabetes mellitus (DM), congestive heart failure (CHF), and pre-existing renal dysfunction (with or without dialysis requirement) were identified using ICD-9 codes listed in Online Table 1. The presence of shock and end-organ failure were identified using ICD-9 codes for cardiogenic shock, noncardiogenic shock, acute renal failure (with or without dialysis or hemofiltration requirement), acute liver failure, and acute respiratory failure in any of the secondary diagnoses fields. Stent type (drug-eluting vs. bare-metal stent), number of stents placed, and use of hemodynamic support during PCI were identified using the appropriate ICD-9 procedure codes. Additionally, PCI-related complications were identified using the appropriate ICD-9 diagnoses and procedure codes in the secondary fields and patient safety indicators proposed by the AHRQ (available: http://www.qualityindicators.ahrq.gov/Modules/PSI_TechSpec.aspx) for post-procedure bleeding and vascular complications. The requirement for blood transfusions (either for bleeding or anemia) was identified using ICD-9 procedure codes (Online Table 1).
Stata SE 13.1 (StataCorp, College Station, Texas) was used for all analyses, accounting for the complex survey design and clustering. For all variables, weighted values of patient-level observations were generated, using prespecified weights in the NIS dataset to produce a nationally representative estimate of the U.S. population of hospitalized patients. Differences between the characteristics of patients undergoing TH versus those not undergoing TH were evaluated using corrected weighted Pearson chi-square test for categorical variables and weighted linear regression or Student t test for continuous variables. Stent thrombosis incidence was computed as the total number of stent thrombosis events divided by the total number of observations. The incidence of stent thrombosis in the TH and the control (i.e., no TH) groups was compared using the corrected weighted Pearson chi-square test. A p value of <0.05 was considered significant.
Multivariable logistic regression models incorporating covariates such as TH, patient demographics, comorbidities, stent type, number of stents, bleeding complications, vascular complications, need for blood transfusions, shock or end-organ failure, use of hemodynamic support, and hospital characteristics were created to determine independent predictors of stent thrombosis. The following variables were included in the model to identify predictors of stent thrombosis: TH, age, sex, hypertension, DM, coronary artery disease, CHF, pre-existing renal dysfunction, type of stent implanted, number of stents, use of hemodynamic support during PCI, bleeding complications, vascular complications, need for blood transfusions, cardiogenic shock, noncardiogenic shock, acute renal failure, acute liver failure, acute respiratory failure, hospital teaching status, hospital size (defined using the number of hospital beds), and hospital region. Additionally, reduced models were generated using the least absolute shrinkage and selection operator method (for low incidence rates), stepwise regression (entry p value of 0.30 and stay p value of 0.35), backward elimination (stay p value of 0.35) and forward selection (entry p value of 0.30) methods. Model fit was assessed using Akaike information criterion and Bayesian information criterion, with lower values indicating better fit.
To account for the baseline differences between patients undergoing TH and those not undergoing TH, propensity score matching was performed. To calculate the propensity score, a logistic regression model was fitted with the following covariates with TH as the outcome: year, age, sex, hypertension, non–insulin-dependent DM, insulin-dependent DM, coronary artery disease, CHF, pre-existing renal dysfunction, type of stent implanted, number of stents, use of hemodynamic support during PCI, bleeding complications, vascular complications, need for blood transfusions, cardiogenic shock, noncardiogenic shock, acute renal failure not requiring dialysis, acute renal failure requiring dialysis or hemofiltration, acute liver failure, acute respiratory failure, hospital teaching status, hospital size, and hospital region. A 1:3 match was then performed based on the propensity score, using a caliper of ±0.01 of the logit of the propensity score. For each matched cohort of patients in the hypothermia and the no hypothermia group, differences in categorical outcomes were assessed by performing Cochran Mantel-Haenszel chi-square test with match-ID as a stratification variable along with survey weights and clusters. For continuous outcomes, linear regression was used with match-ID as a stratification variable along with survey weights and clusters. The incidence of stent thrombosis between the 2 groups after propensity matching was compared using the Cochran Mantel-Haenszel test.
From 2006 to 2011, there were 49,109 CA patients with AMI undergoing PCI. Of these, 1,193 (2.4%) underwent TH. Table 1 shows the baseline characteristics of the study population in the TH and the no TH groups.
Stent thrombosis incidence
There were 2,317 stent thrombosis events (4.7%), of which 46 (3.9%) occurred in the TH group (n = 1,193) and 2,271 (4.7%) occurred in the no TH group (n = 47,916). There was no difference in the stent thrombosis incidence between the TH and the no TH groups (3.9% vs. 4.7%; p = 0.61). There was a nonsignificant trend toward an increase in stent thrombosis incidence with age, with stent thrombosis incidence in age groups of 18 to 40 years, 41 to 50 years, 51 to 60 years, 61 to 70 years, 71 to 80 years and >80 years being 2.4%, 4.1%, 4.5%, 4.8%, 5.4%, and 5.4%, respectively (p = 0.13). The stent thrombosis incidence in those who received drug-eluting stent was not different from those who received a bare-metal stent (4.8% vs. 4.6%; p = 0.70). The incidence of stent thrombosis in those who received hemodynamic support was significantly higher than those who did not (6.3% vs. 4.1%; p < 0.001). The incidence of stent thrombosis increased with increase in the number of stents implanted, with the stent thrombosis incidence for 1, 2, 3, and ≥4 stent implantations being 4.1%, 5.3%, 6.2%, and 7.9%, respectively (p < 0.001). The incidence of stent thrombosis was greater in those who had CHF (5.8%) compared with those who did not have CHF (4.2%; p = 0.001). Similarly, stent thrombosis incidence was higher in the presence of cardiogenic shock (6.2% vs. 4.0%; p < 0.001). The incidence of stent thrombosis was higher in those who had a bleeding complication (5.9% vs. 4.5%; p = 0.03) or a vascular complication (8.8% vs. 4.6%; p = 0.002); and in those who needed blood transfusions (6.3% vs. 4.6%; p = 0.02).
Predictors of stent thrombosis
Logistic regression analysis showed that TH was not a significant predictor of stent thrombosis. The adjusted odds ratio (OR) of TH in predicting stent thrombosis was 0.7 (95% confidence interval [CI]: 0.3 to 1.8; p = 0.46). Age >80 years (OR: 2.1; 95% CI: 1.0 to 4.4; p = 0.05), CHF (OR: 1.2; 95% CI: 1.0 to 1.5; p = 0.05), use of hemodynamic support (OR: 1.3; 95% CI: 1.0 to 1.6; p = 0.04), occurrence of vascular complications (OR: 1.7; 95% CI: 1.1 to 2.6; p = 0.02), and presence of cardiogenic shock (OR: 1.3; 95% CI: 1.0 to 1.6; p = 0.04) were independent predictors of stent thrombosis (Table 2). Additionally, there was a progressive increase in the risk of stent thrombosis with increasing number of stents implanted: 2 stents (OR: 1.3; p = 0.03), 3 stents (OR: 1.4; p = 0.07), and ≥4 stents (OR: 1.8; p = 0.01) (Table 2).
Reduced models generated using the least absolute shrinkage and selection operator method and stepwise regression, forward selection, and backward elimination methods are shown in Online Tables 2 and 3, respectively. Notably, stepwise, forward and backward regression methods yielded the exact same variables; therefore, the same reduced model was generated with these 3 methods (Online Table 3). The Akaike information criterion and Bayesian information criterion values of the full model were not different from those of the reduced models, and were in fact slightly lower with the full model. Moreover, the odds ratios and p values of important predictors such as TH, number of stents, hemodynamic support use, vascular complications, CHF, and cardiogenic shock were similar in the full and reduced models (Table 2, Online Tables 2 and 3). Therefore, only the results of the full model are presented.
Using propensity score analysis, we were able to match 1,155 patients in the TH group with 3,399 patients in the no TH group. Table 3 shows that the baseline characteristics in the TH and no TH groups were well-balanced after propensity matching. There was no difference in the incidence of stent thrombosis in the TH group (3.6%) compared with the no TH group (6.1%; p = 0.17) after propensity matching.
Overall, the in-hospital mortality rate was 26.3%. The in-hospital mortality rates did not differ significantly in those who experienced stent thrombosis (24.8%) compared with those who did not experience stent thrombosis (26.4%; p = 0.42).
Length of stay
Overall, the average mortality-free LOS was 8.8 days. In the overall population, the average mortality-free LOS was higher in the TH group compared with the no TH group (13.0 ± 10.2 days vs. 8.7 ± 8.9 days; p < 0.001), whereas the average LOS in those who died was similar between the 2 groups (4.7 vs. 4.5 days; p = 0.74). After propensity matching, there was no difference in the average mortality-free LOS (12.3 vs. 10.4 days; p = 0.06) or mortality-related LOS (4.7 vs. 4.2 days; p = 0.39) between those undergoing TH compared with those not undergoing TH.
To the best of our knowledge, this is the largest study to compare the incidence of stent thrombosis in patients with CA undergoing TH to those not undergoing TH. We performed a retrospective analysis on a large in-patient population of CA patients admitted across the country from 2006 to 2011. Major findings of our study are as follows: 1) The incidence of stent thrombosis after PCI in AMI patients presenting with CA was 4.7%; and 2) TH did not increase the incidence of stent thrombosis in CA patients with AMI undergoing PCI.
Clinical trial and registry data have demonstrated stent thrombosis rates after PCI to be <1% (19). The incidence of stent thrombosis may be higher in the setting of acute coronary syndromes (ACS) (20–22). The HORIZONS-AMI (Harmonizing Outcomes With Revascularization and Stents in Acute Myocardial Infarction) trial reported an 0.8% incidence of acute (within 24 h) definite or probable stent thrombosis and 1.4% incidence of subacute (24 h to 30 days) stent thrombosis in patients undergoing primary PCI for AMI (22,23). Other studies have demonstrated similar findings in AMI patients not presenting as OHCA. Penela et al. (12) reported an 0.7% rate of definite stent thrombosis after primary PCI, whereas Rosillo et al. (14) reported a 1.2% incidence of acute stent thrombosis and a 1% incidence of subacute stent thrombosis in patients undergoing primary PCI for STEMI. However, the incidence of stent thrombosis in the setting of CA has been poorly studied.
There is a broad range of reported incidence of stent thrombosis in CA patients in the literature, ranging from 1.4% to 45.5% (Table 4) in various studies (11,12,14,15,24,25). Consistent with prior observations of a higher incidence of stent thrombosis in CA, we observed a 4.7% in-hospital stent thrombosis incidence in our population of CA patients. Because the NIS data pertain to a single hospitalization, all stent thrombosis events in our study represent acute and subacute stent thrombosis only. The incidence of acute and subacute stent thrombosis in our study (4.7%) was more than twice that reported in the HORIZONS-AMI trial (2.2%) (23), further affirming a higher stent thrombosis risk in the CA population.
Our study also identified several independent predictors of acute and subacute stent thrombosis in the CA population. Advancing age (>80 years), presence of CHF or cardiogenic shock, and increasing number of stent implantations resulted in a greater risk of stent thrombosis. It is reassuring that there was no association between the type of stent (bare metal vs. drug eluting) and risk of stent thrombosis. Interestingly, we found a higher incidence of stent thrombosis in patients receiving hemodynamic support (OR: 1.3; p = 0.04), in contrast with the conventional belief that hemodynamic support has a protective effect on stent thrombosis. We propose that the need for hemodynamic support is actually a marker of critical illness and circulatory collapse in our patient population, and that the circulatory collapse, rather than the use of hemodynamic support, is responsible for the greater risk of stent thrombosis in this subgroup. In an unadjusted analysis, we observed that the occurrence of bleeding or vascular complications and the need for blood transfusions (due to bleeding or anemia) was associated with a higher incidence of stent thrombosis. The occurrence of anemia, bleeding, or vascular complications often leads to interruption or discontinuation of dual antiplatelet therapy after PCI, which in turn pre-disposes to a higher risk of stent thrombosis (26). In multivariate analysis, only the occurrence of vascular complications (OR: 1.7; p = 0.02) remained a significant predictor of stent thrombosis. Of note, we also observed that stent thrombosis did not increase in-hospital mortality in our patient population (24.8% with stent thrombosis vs. 26.4% without stent thrombosis; p = 0.41). It is unclear why stent thrombosis was not associated with an increased mortality risk; however, these findings are consistent with prior observations of better than expected survival after early stent thrombosis in the setting of CA (15,27).
Several mechanisms have been proposed to explain the higher incidence of stent thrombosis in CA patients. The comatose state of patients after resuscitation makes it difficult to administer oral medications, which results in an inability to load these patients with dual antiplatelet therapy prior to cardiac catheterization. Joffre et al. (15) in their population of AMI patients undergoing PCI showed that none of the patients who presented with CA received dual antiplatelet therapy before PCI. Administration of antiplatelet agents via orogastric or nasogastric tubes after crushing the medications may result in unreliable gastrointestinal absorption of these drugs. Additionally, shock and multiorgan failure after CA can result in delayed absorption and metabolism of antiplatelet drugs (28), some of which are prodrugs (e.g., clopidogrel and prasugrel) and need to be metabolized by the liver to be converted to an active metabolite. Additionally, the critical illness of the CA patients can result in a hypercoagulable and prothrombotic state (12,29). The residual effect of high doses of vasoconstrictor medications administered during cardiopulmonary resuscitation and a high level of circulating catecholamines and inflammatory cytokines in the immediate postresuscitation period may also play an important role. Other proposed mechanisms to explain the greater risk of stent thrombosis include no flow followed by restoration of flow associated with successful resuscitation, and associated myocardial microcirculatory and endothelial dysfunction (30,31).
The relationship between TH and stent thrombosis remains controversial, with a major limitation of inadequate power or sample size in the studies evaluating this association. The current literature and their major findings on this topic are summarized below. In 2011, Ibrahim (11) reported a 14.8% stent thrombosis incidence in 27 CA patients undergoing TH and PCI, compared with no stent thrombosis events in 30 CA patients undergoing PCI but no TH. In 2013, Penela et al. (12) observed 5 stent thrombosis events in 11 CA patients treated with mild TH, and concluded that the risk of stent thrombosis is significantly increased in patients undergoing TH. Kozinski et al. (13) showed no stent thrombosis events in 37 OHCA patients with ACS undergoing TH and PCI. In 2014, Rosillo et al. (14) reported a 2.7% incidence of stent thrombosis in 77 patients undergoing TH and primary PCI, which was not different from a 2.3% stent thrombosis incidence in 1,414 patients undergoing primary PCI for STEMI during the same time period. Casella et al. (7) also showed no stent thrombosis events in 45 CA patients undergoing PCI and treatment with TH. Joffre et al. in 2014 (15) and Gouffran et al. in 2015 (25) observed a 10.9% stent thrombosis incidence in their population of respectively 55 and 101 patients undergoing TH after primary PCI. In 2015, Erlinge et al. (32) in a combined analyses of 2 prospective randomized controlled trials (RAPID MI-ICE [Rapid Intravascular Cooling in Myocardial Infarction as Adjunctive to Percutaneous Coronary Intervention] and CHILL-MI [Efficacy of Endovascular Catheter Cooling Combined With Cold Saline for the Treatment of Acute Myocardial Infarction]) with a total of 140 patients (70 in each group), reported only 1 patient in the TH group to have a re-infarction; stent thrombosis was not mentioned specifically (32). Finally, a recent meta-analysis by Villablanca et al. (33), including data from 6 randomized controlled trials (total 819 patients) to evaluate the efficacy and safety of TH in STEMI patients, showed no difference in all-cause mortality, major adverse cardiovascular events, or recurrent infarction between those who received TH and those who did not.
The mechanisms proposed to explain increased risk of stent thrombosis with TH include impaired metabolism and bioavailability of antiplatelet drugs, reduced or ineffective platelet inhibition in the presence of TH (34–36), TH-related enhanced thrombogenicity, increased platelet activation, reduced adenosine diphosphate clearance, increased shedding of platelet microparticles, and TH-induced mast cell degranulation (10,37–42). Indirect measures, such as the platelet reactivity index (PRI), have been explored to evaluate the changes in platelet function associated with TH (43). High PRI values are associated with a higher incidence of arterial thromboembolic complications (44). In the ISAR SHOCK (Efficacy Study of LV Assist Device to Treat Patients With Cardiogenic Shock) registry, Orban et al. (45) examined platelet reactivity and clinical outcomes (including stent thrombosis) in patients with ACS complicated by cardiogenic shock undergoing PCI. Sixty-four patients who received TH were compared with 81 patients who did not receive TH. TH use was not associated with any change in platelet reactivity, 30-day mortality, or re-infarction. Only 3 definite stent thrombosis events were identified, all in the TH group (5% vs. 0%; p = 0.09) (45). Ibrahim et al (35) examined PRI in patients with ACS and CA, divided into those undergoing TH (n = 84) and those not undergoing TH (n = 80). There was a significantly higher PRI (and thereby less platelet inhibition) in patients undergoing TH. The reduction in platelet inhibitory effect with TH was seen with all 3 antiplatelet drugs (clopidogrel, prasugrel, and ticagrelor); however, it was most marked in patients receiving clopidogrel (35).
It is evident that most studies looking at the relationship between TH and stent thrombosis are small, single-center studies, with a limited number of patients in the TH group, ranging from 11 to 101 (Table 4). Owing to the small number of TH patients in these studies, there is a possibility that the higher incidence of stent thrombosis with TH may be due to chance (27). Moreover, most studies on TH and stent thrombosis either lack a control group or have non-CA AMI patients as their control group. As discussed earlier, the post-CA state itself can pre-dispose the patients to a higher risk of stent thrombosis. Therefore, in order to determine the true contribution of hypothermia to the occurrence of stent thrombosis, it is important to compare the stent thrombosis rates in CA AMI patients undergoing TH with those in CA AMI patients not undergoing TH. Our study is the largest reported study focusing on the relationship between TH and stent thrombosis, including 1,193 TH patients from multiple hospitals across the United States, and using CA AMI patients not undergoing TH as the comparison group. We observed that, although the incidence of acute and subacute stent thrombosis in CA AMI patients is high (4.7%), it is not higher in those undergoing TH (3.9%) compared to those not undergoing TH (4.7%; p = 0.16). These findings persist after propensity matching to account for baseline differences between the TH and no TH groups.
Being an administrative database relying on ICD-9 codes, the NIS is susceptible to errors arising from coding inaccuracies. It is not possible to track patients after being discharged from the hospital. Therefore, all data are cross-sectional and the event rates (including stent thrombosis) reported refer to index hospitalization only. Long-term outcomes (e.g., long-term stent thrombosis rates) cannot be assessed from this study. Of note, only 1,193 of 49,019 CA patients (2.4%) in our study population received TH. The reason behind this may be the inability to precisely identify CA patients eligible to receive TH due to the administrative nature of the database. Hence, our denominator includes all CA patients, and not just the CA patients who would be eligible to receive hypothermia (e.g., those who continue to remain unconscious after resuscitation). Therefore, the TH rate in our study is clearly an underestimation; however, our study was not intended at evaluating the TH rates. Stent thrombosis was identified using ICD-9 diagnosis code 996.72; however, there is no information on whether the stent thrombosis events were definite, probable, or possible stent thrombosis. There is no information about medication use in the NIS database; hence, it is not possible to determine which patients received dual antiplatelet therapy before the PCI, and what antithrombotic medications were administered during or after the PCI (e.g., heparin, bivalirudin, glycoprotein IIb/IIIa inhibitors). There is no information in the NIS database about angiographic features such as thrombus burden, stent size, TIMI (Thrombolysis In Myocardial Infarction) flow post stent, any residual disease proximal or distal to the stent, stent edge dissection, stent underexpansion or asymmetric expansion, stent malapposition, or final luminal diameter after stent placement, all of which have been shown to be predictors of stent thrombosis (19).
There is a high incidence of stent thrombosis after primary PCI in AMI patients presenting with CA (4.7%). Using a large population-based retrospective analysis, we show that the use of TH does not increase the risk of stent thrombosis in these patients.
The role of newer purinergic receptor G-protein coupled 12 (P2Y12) inhibitors (prasugrel and ticagrelor) in the setting of CA is unclear and controversial, with some data showing improved platelet inhibition (35), but more recent data showing surprisingly higher rates of stent thrombosis compared with clopidogrel (25). The use of intravenous P2Y12 inhibitors such as cangrelor can result in better platelet inhibition, especially in the setting of CA; however, it remains to be seen whether this translates into a reduction in stent thrombosis events. Future prospective studies and randomized controlled trials are needed to explore the role of these medications in AMI patients presenting with CA.
WHAT IS KNOWN? Theoretically, alterations in platelet reactivity and pharmacokinetics of antiplatelet agents with TH can pre-dispose resuscitated CA patients undergoing concomitant TH and PCI to a higher risk of acute and subacute stent thrombosis. However, the current literature contains only a handful of small studies on this topic, which show an inconsistent relationship between TH and stent thrombosis.
WHAT IS NEW? Our large nationwide study including 49,109 patients with CA due to AMI admitted to hospitals across the United States from 2006 to 2011 showed that the overall incidence of stent thrombosis after PCI in these patients was high (4.7%). However, the use of TH did not increase the risk of stent thrombosis, indicating that TH can be safely combined with primary PCI in AMI patients presenting as cardiac arrest.
WHAT IS NEXT? The role of newer antiplatelet agents (prasugrel, ticagrelor, and cangrelor) to reduce the risk of acute and subacute stent thrombosis in the setting of CA needs to be explored.
For supplemental tables, please see the online version of this article.
Dr. Cox has reported that he is a member of the medical advisory board for Abbott Vascular, The Medicines Company, and Boston Scientific. Dr. Kern is a member of the Science Advisory Board Member for Zoll Medical.
- Abbreviations and Acronyms
- acute coronary syndrome(s)
- acute myocardial infarction
- cardiac arrest
- congestive heart failure
- confidence interval
- International Classification of Diseases
- diabetes mellitus
- length of stay
- Nationwide Inpatient Sample
- out-of-hospital cardiac arrest
- odds ratio
- percutaneous coronary intervention
- platelet reactivity index
- ST-segment elevation myocardial infarction
- therapeutic hypothermia
- Received May 12, 2016.
- Accepted June 2, 2016.
- 2016 American College of Cardiology Foundation
- Mozaffarian D.,
- Benjamin E.J.,
- Go A.S.,
- et al.
- Fugate J.E.,
- Brinjikji W.,
- Mandrekar J.N.,
- et al.
- Dumas F.,
- Cariou A.,
- Manzo-Silberman S.,
- et al.
- Neumar R.W.,
- Shuster M.,
- Callaway C.W.,
- et al.
- Straub A.,
- Krajewski S.,
- Hohmann J.D.,
- et al.
- Ibrahim K.
- Penela D.,
- Magaldi M.,
- Fontanals J.,
- et al.
- Rosillo S.O.,
- Lopez-de-Sa E.,
- Iniesta A.M.,
- et al.
- Byrne R.A.,
- Joner M.,
- Kastrati A.
- Kukreja N.,
- Onuma Y.,
- Garcia-Garcia H.M.,
- Daemen J.,
- van Domburg R.,
- Serruys P.W.
- Brodie B.,
- Pokharel Y.,
- Garg A.,
- et al.
- Dangas G.D.,
- Caixeta A.,
- Mehran R.,
- et al.
- Urban P.,
- Abizaid A.,
- Banning A.,
- et al.
- Adrie C.,
- Monchi M.,
- Laurent I.,
- et al.
- Brenoche C.S.M.,
- Timerman S.,
- Kopel L.,
- Kern K.B.,
- Lage S.
- Villablanca P.A.,
- Rao G.,
- Briceno D.F.,
- et al.
- Michelson A.D.,
- MacGregor H.,
- Barnard M.R.,
- Kestin A.S.,
- Rohrer M.J.,
- Valeri C.R.
- Koscielny J.,
- Aslan T.,
- Meyer O.,
- et al.
- Orban M.,
- Mayer K.,
- Morath T.,
- et al.