Major bleeding complications after cardiopulmonary resuscitation: impact of thrombolytic treatment



Abstract.  Kurkciyan I, Meron G, Sterz F, Müllner M, Tobler K, Domanovits H, Schreiber W, Bankl HC, Laggner AN (General Hospital of Vienna, University of Vienna, Vienna, Austria). Major bleeding complications after cardiopulmonary resuscitation: impact of thrombolytic treatment. J Intern Med 2003; 253: 128–135.

Objective.  The risk of bleeding complications caused by thrombolysis in patients with cardiac arrest and prolonged cardiopulmonary resuscitation is unclear. We evaluate the complication rate of systemic thrombolysis in patients with out-of-hospital cardiac arrest caused by acute myocardial infarction, especially in relation to duration of cardiopulmonary resuscitation.

Design.  The study was designed as retrospective cohort study, the risk factor being systemic thrombolysis and the end-point major haemorrhage, defined as life-threatening and/or need for transfusion. Over 10.5 years, emergency cardiac care data, therapy, major haemorrhage and outcome of 265 patients with acute myocardial infarction admitted to an emergency department after successful cardiopulmonary resuscitation were registered.

Results.  We observed major haemorrhage in 13 of 132 patients who received thrombolysis (10%, 95% confidence interval 5–15%), five of these survived to discharge, none died because of this complication. Major haemorrhage occurred in seven of 133 patients in whom no thrombolytic treatment had been given (5%, 95% confidence interval 1–9%), two of these survived to discharge. Taking into account baseline imbalances between the groups, the risk of bleeding was slightly increased if thrombolytics were used (odds ratio 2.5, 95% confidence interval 0.9–7.4) but this was not significant (P = 0.09). There was no clear association between duration of resuscitation and bleeding complications (z for trend = 1.52, P = 0.12). Survival was not significantly better in patients receiving thrombolysis (odds ratio 1.6, 0.9–3.0, P = 0.12).

Conclusions.  Bleeding complications after cardiopulmonary resuscitation are frequent, particularly in patients with thrombolytic treatment, but do not appear to be related to the duration of resuscitation. In the light of possible benefits on outcome, thrombolytic treatment should not be withheld in carefully selected patients.


In patients with acute myocardial infarction (AMI), cardiac arrest requiring cardiopulmonary resuscitation (CPR) is a common complication. For AMI, systemic thrombolytic treatment is well established. Numerous large clinical trials of thrombolytic therapy have shown impressive reductions in size of infarction, complications and, finally, mortality [1–5]. However, deciding whether to use thrombolytic treatment for patients who have suffered cardiac arrest and received CPR as a result of AMI is difficult, because the already significant risk of haemorrhage related to thrombolysis might be increased by CPR. External chest compression during resuscitation is frequently associated with serious complications of varying severity, especially bleeding complications [6]. In such cases, thrombolysis may increase the severity of bleeding complications following chest compression. For this reason, the 1999 guidelines of the American College of Cardiology/American Heart Association for treatment of AMI still list prolonged CPR (>10 min) as a relative contraindication against thrombolytic therapy [7–10]. Many major clinical trials on thrombolysis in AMI have intentionally excluded these patients [2, 10–12].

Only some case reports and small case series provided data regarding the safety and efficacy of systemic thrombolytic treatment in patients who suffered cardiac arrest caused by AMI [13–20]. Some authors consider CPR to be a relative contraindication for thrombolysis, irrespective of the duration of CPR, whilst others consider only prolonged CPR a contraindication.

Aim of our study was to describe major bleeding complications in a large cohort of resuscitated patients with out-of-hospital cardiac arrest caused by AMI in order to evaluate the safety of systemic thrombolysis in these patients, especially in relation to the duration of CPR.

Materials and methods

From July 1991 to December 2001, all patients admitted to the emergency department of an urban tertiary care University hospital after out-of-hospital cardiac arrest were documented according to a specific protocol (Utstein Style) [21]. All patients admitted after witnessed cardiac arrest and diagnosis of AMI were included in the study. In a recently published study [22] 144 of our 265 patients have been described. The aim of this study [22] was to describe the association between thrombolytic therapy and neurological recovery, the inclusion criteria were different, and the observational period for the present study is almost twice as long.

We documented the following data on arrival: sex and age of the patient, the presumed cause of cardiac arrest, initial electrocardiogram (ECG) rhythm, no-flow time, low-flow time, and outcome. No-flow time was defined as estimated interval from collapse to first cardiopulmonary resuscitation attempts. We defined low-flow time as the time interval from the begin of basic or advanced life support measures until return of spontaneous circulation. It serves as a measure for the duration of trauma to the chest. We assessed cerebral function prospectively on arrival and at regular intervals within 6 months after return of spontaneous circulation in terms of cerebral performance categories (CPC 1–5) [23, 24]. Cerebral Performance Categories 1 and 2 were defined as a good, CPC 3–5 as an unfavourable neurological recovery. According to the criteria of the American Heart Association, we defined cardiac arrest as sudden collapse followed by loss of consciousness and absence of both spontaneous respiration and pulse. Therapy for cardiac arrest was in accordance with international guidelines [25, 26]. Most patients received standard intensive care treatment with mechanical ventilation, sedation, analgesia and stress ulcer prophylaxis.

Diagnosis of AMI was based on 12-lead ECG's showing ST-segment elevation >0.2 mV in two precordial leads or >0.1 mV in two limb leads and subsequent development of Q waves within the hospital stay. In nonsurvivors, the diagnosis of AMI was confirmed by autopsy in all cases. We reviewed the clinical records of all primarily successfully resuscitated patients admitted after nontraumatic out-of-hospital cardiac arrest with definitive diagnosis of AMI. In these patients, the history, ECG, the site of AMI, specific therapy, major bleeding complications and outcome were evaluated.

The decision for using thrombolytic treatment was made by the admitting emergency physician when AMI was diagnosed as the immediate cause of cardiac arrest. In the presence of contraindications to thrombolytic therapy, known underlying severe morbidity and expected neurological damage after prolonged resuscitation efforts, thrombolysis was not performed. If thrombolytic therapy was given, 100 mg of recombinant tissue-plasminogen activator (rt-PA) was administered. The thrombolytic agent alteplase (Actilyse®; Boehringer Ingelheim, Biberach, Germany) was given as a bolus dose of 15 mg, followed by continuous infusion of 50 mg over 30 min and 35 mg over the next 60 min. All patients undergoing thrombolysis received aspirin and a bolus of 4000–5000 U followed by a continuous infusion of 1000 U of heparin per hour. Heparin was adjusted at 6-h intervals.

We analysed the clinical course of each patient for the following major bleeding complications: (i) bleeding in body cavities: haematopericardium, haematothorax, haemoperitoneum; (ii) intraparenchymal haematoma following organ rupture or laceration; (iii) intracranial bleeding; (iv) any bleeding necessitating transfusion of packed red blood cells.

We used a retrospective cohort study design to assess the association between thrombolysis (the risk factor) and major bleeding complications (the primary end-point) and survival (the secondary end-point).

Statistical methods

Continuous data are given as median and the interquartile range; frequencies are given as counts and percentages. We compare groups of continuous data with the Mann–Whitney U-test. Groups of categorical data are compared by means of χ2 test, or Fisher's exact test, if appropriate. Trends across ordered groups are measured with a test developed by Cuzick, an extension of the Wilcoxon rank-sum test, or an exact χ2 test for trend [27]. Confidence intervals were calculated for outcome estimates.

We stratified for the duration of trauma to the chest (quartiles of low-flow time) to assess the association between thrombolysis and bleeding complications, whilst controlling for this potential confounder. Stratification also allows to assess a possible interaction between thrombolysis and duration of CPR. We used quartiles of the variable low-flow time because this allows groups of equal size. For the quantitative combination of stratified binary variables the Mantel–Haenszel odds ratio was used.

We used binary logistic regression to assess the association between the exposure (thrombolytic treatment) and the end-point (major bleeding complications) whilst controlling for several baseline imbalances. Variables which were unequally distributed between the groups indicated by P < 0.1 were included in the model. Regression diagnostics were performed according to standard methods [28]. The overall model fit was assessed with the Hosmer–Lemeshow test, with a P-value >0.1 indicating an acceptable fit. Calculations were performed with Stata release 7, or with StaXact version 5.


During the 10.5-year period, 859 patients were admitted to the emergency department after successful resuscitation following out-of-hospital cardiac arrest. Acute myocardial infarction was identified as the immediate cause of cardiac arrest in 313 (36%) patients. Forty-eight patients treated with acute angioplasty as first line therapy were subsequently excluded from the analysis. Of the remaining 265 patients, 132 (50%) were in the thrombolysis group and 133 (50%) in the nonthrombolysis group. Overall, major bleeding complications were observed in 20 (6%) of the 313 patients.

Patients with major bleeding complications

In the thrombolysis group we observed major bleeding complications (Table 1) in 13 of 132 patients (10%, 95% confidence interval 5–15%) (patient nos 1–13). In four patients (1, 2, 7 and 8) bleeding at the puncture site of central venous access and epistaxis occurred, requiring transfusion of packed red blood cells. In patient 3 (CPR duration 13 min), intracerebral haemorrhage was found by cranial computed tomography. At the time of medical decision-making it was not known that this patient had a history of chronic heart failure and was on coumadin therapy. Patient 4 (CPR duration 33 min) died 13 days after admission and thrombolysis. At autopsy intracerebral haemorrhage in the region of the basal ganglia was found. Patients 5 and 6 developed severe gastrointestinal haemorrhage following thrombolysis. They could be stabilized after transfusion of coagulation factors and red blood cells. In patient 9 (CPR duration 21 min) rupture of the left liver lobe as well as rupture of the spleen accompanied by haemoperitoneum was diagnosed by sonography and computer tomography of the abdomen after thrombolysis. Urgent laparotomy was performed. The patient recovered without neurological damage to be discharged after 24 days. In patients 10 (CPR duration 16 min) and 11 (CPR duration 29 min), haematoma of the upper extremities, thorax and shoulder region were noted after thrombolysis and required transfusion of packed red blood cells. Patient 12 (CPR duration 3 min) developed upper gastrointestinal bleeding post thrombolysis which resolved after substitution of two packed red blood cells and proton pump inhibitor therapy. Patient 13 (CPR duration 33 min) developed multiorgan-failure and died within 2 days after admission. Autopsy findings included myocardial bleeding in the inferior septum as well as bleeding in the gastric mucosa.

Table 1.  Characteristics and major bleeding complications of resuscitated patients with and without thrombolysis
Patient no./
Sex/age, years
GroupDuration of
CPR, min
Units of red
blood cells
  1. CPR = Cardiopulmonary resuscitation; M = male; F = female.

1/F/73Thrombolysis7Epistaxis, bleeding at the puncture site8Died, 93 days
2/F/60Thrombolysis4Bleeding at the puncture sites4Died, 8 days
3/M/58Thrombolysis13Cerebral haemorrhage0Died, 11 days
4/M/54Thrombolysis33Cerebral haemorrhage0Died, 13 days
5/M/33Thrombolysis26Gastrointestinal haemorrhage8Died, 33 days
6/F/68Thrombolysis12Gastrointestinal haemorrhage, epistaxis6Alive
7/M/46Thrombolysis58Epistaxis2Died, 1 day
8/F/43Thrombolysis22Bleeding at the puncture sites2Alive
9/M/52Thrombolysis21Rupture of liver and spleen, haemoperitoneum6Alive
10/M/60Thrombolysis16Haemorrhagic suffusions in the knee and thorax2Alive
11/M/64Thrombolysis29Haemorrhagic suffusions in the extremities2Died, 4 days
12/F/72Thrombolysis3Gastrointestinal haemorrhage2Alive
13/M/50Thrombolysis33Septum haemorrhage, intestinal haemorrhage0Died, 2 days
14/M/83Nonthrombolysis12Haemato-pneumothorax4Died, 1 day
15/M/66Nonthrombolysis75Rupture of liver, haemoperitoneum4Died, 1 day
16/M/49Nonthrombolysis20haematoma of liver0Died, 7 days
17/M/50Nonthrombolysis18Septum haemorrhage0Died, 7 days
18/F/76Nonthrombolysis19Haemato-pneumothorax4Died, 1 day
20/M/50Nonthrombolysis13Cerebral haemorrhage0Alive

In the nonthrombolysis group we observed major bleeding complications in seven of 133 patients (5%, 95% confidence interval 1–9%) (patient nos 14–20). In patient 14 (CPR duration 12 min) the X-ray revealed left-sided fluidopneumothorax without signs of tension as major complication. A chest tube was inserted, and an intracavital haemorrhage was proven. In patient 15 (CPR duration 75 min) ultrasound examination showed a laceration of the liver capsule with free fluid within the peritoneal cavity; peritoneal drainage revealed haemoperitoneum, and the patient died prior to surgical intervention. In patient 16 (CPR duration 20 min) who died 7 days after the index event, a capsular tear in the left liver lobe coupled with intraparenchymal haematomas was found at autopsy. In patient 17 (CPR duration 18 min) a large haemorrhage in the cardiac intraventricular septum was found at autopsy. In patient 18 (CPR duration 19 min) transthoracic echocardiograpy as well as X-ray and autopsy showed a haemato-pneumothorax. In patient 19 (CPR duration 23 min) a routine chest X-ray on day 3 after admission showed unilateral fluido-haematothorax caused by serial rib fractures, requiring chest drain. Patient 20 (CPR duration 13 min) had suffered a fracture of the occiput with a 6 cm diameter contusion bleeding and intraventricular haemorrhage as well as serial rib fractures. Both patients 19 and 20 survived to discharge.

Association between thrombolysis and major bleeding complications

Patients in the thrombolysis group were younger, had a shorter duration of trauma to the chest (low-flow time) and no-flow time, had more often ventricular fibrillation and anterior site of AMI in the first ECG and had less often a history of myocardial infarction. Survival over 6 months was significantly better in patients receiving thrombolysis compared with patients who did not receive thrombolysis (Table 2). The duration of CPR, given in quartiles, was not associated with major bleeding complications (z for trend = 1.52, P = 0.41), irrespective of thrombolytic treatment (Table 3).

Table 2.  Comparison of the thrombolysis with the nonthrombolysis group for common factors
FactorsThrombolysis group
(n = 132)
Nonthrombolysis group
(n = 133)
  1. Values are expressed as n (%) or median (range). F = female; AMI = Acute myocardial infarction.

Sex, F29 (22)29 (22)0.974
Age, years55 (47–64)62 (54–71)0.001
Previous myocardial infarction23 (17)38 (29)0.031
No-flow time, min1 (0–5)3 (0–9)0.003
Low-flow time, min12 (4–24)17 (10–31)0.001
Anterior site of AMI79 (60)58 (44)0.008
Ventricular fibrillation119 (90)101 (76)0.002
Major bleeding complications13 (10)7 (5)0.158
Survival83 (63)47 (35)<0.001
Cerebral performance category 1/268 (52)37 (28)<0.001
 (of alive patients) after 6 months
Table 3.  Association between thrombolysis and major bleeding complications according to the duration of cardiopulmonary resuscitation
Time quartilesTotal no of
Bleeding complications in
thrombolysis group, no (%)**
Bleeding complications in
nonthrombolysis group, no (%)**
  • *

    Exact test for trend, P = 0.16.

  • **

    Exact test for trend, P = 0.12.

Low-flow 1–6 min2/672/460/21
Low-flow 7–14 min4/603/241/36
Low-flow 15–29 min9/595/264/33
Low-flow >29 min5/593/232/36

Patients who underwent systemic thrombolysis subsequently had slightly more often bleeding complications (unadjusted odds ratio 2.0, 95% confidence interval 0.8–5.1, P = 0.16). This effect was, however, statistically not significant. When stratifying for increasing duration of CPR (quartiles: 1–6, 7–14, 15–29 and >29 min) the risk of a bleeding complication was slightly increased. (Mantel–Haenszel odds ratio 2.4, 95% confidence interval 0.9–6.7, P = 0.07; χ2 for heterogeneity = 0.98, d.f. = 3, P = 0.80).

When adjusting for duration of CPR (quartiles) and all other above mentioned baseline imbalances [age (quartiles), no-flow time (quartiles), ventricular fibrillation in the first ECG, location of myocardial infarction (anterior versus posterior) and history of myocardial infarction (yes versus no)], the odds ratio for bleeding complications in patients with thrombolysis remained unchanged (2,5, 95% confidence interval 0.9–7.4, P = 0.09) compared with patients who did not receive thrombolysis. The model had an acceptable fit (Hosmer–Lemeshow χ2 = 2.73, d.f. = 8, P = 0.95).

Association between thrombolysis and survival

Survival over 6 months was better in patients receiving thrombolysis compared with patients who did not receive thrombolysis (odds ratio 3.1, 95% confidence interval 1.8–5.2, P < 0.001). Patients receiving thrombolysis were younger, had less often prior myocardial infarction, had shorter duration of no-flow and low-flow, and more often ventricular fibrillation (Table 2). When adjusting for these baseline imbalances the effect was greatly reduced (odds ratio 1.6, 95% confidence interval 0.88–3.0, P = 0.12) (Hosmer–Lemeshow χ2 = 6.37, d.f. = 8, P = 0.60).


In a large cohort of patients with out-of-hospital cardiac arrest as a result of acute myocardial infarction, we found major bleeding complications in 10% of 132 patients with thrombolysis and CPR, almost half of them survived to discharge. None of our patients died because of a CPR-related bleeding complication. Major haemorrhage occurred in 5% of 133 patients in whom no thrombolytic treatment had been given, about a quarter of them survived to discharge. Risk of bleeding was higher if thrombolytics were used. There was, however, no clear association between increasing duration of CPR and major bleeding complications.

There is a consensus that thrombolytic therapy may safely be performed if the duration of CPR is <10 min [8, 9]. In a retrospective analysis, Tenaglia et al. [18] studied 59 patients after short duration of CPR (<10 min), 37 of whom required only defibrillation, excluding patients with a resuscitation procedure lasting more than 10 min They found no direct complications of thrombolytic therapy and concluded that thrombolytic treatment after short duration of CPR is safe and effective. Weston and Avery [20] found three significant bleeding complications after thrombolysis in a group of 16 patients, all of these patients survived to hospital discharge. The authors recommended to perform thrombolysis only in patients with overwhelming electrocardiographic evidence of AMI, and even then with caution.

Whether thrombolytic treatment should be administered in patients in whom CPR was performed for more than 10 min is still controversial. Scholz et al. [17] found no bleeding complications directly related to CPR in 16 patients after successful resuscitation – but in only four of them systemic thrombolysis was performed. Van Campen et al. [19] also evaluated the safety of thrombolytic therapy in a group of 33 patients who were treated with thrombolytics after successful CPR. Except for one patient who died of hypovolemic shock after gastrointestinal bleeding, no severe bleeding complications occurred. Based on this small sample, the authors stated that prolonged CPR (>20 min) by itself should not be considered a contraindication to thrombolytic therapy. Jäger et al. [15] and Cross et al. [14] found no bleeding complications in a small sample of patients (n = 11 and 10, respectively) who received CPR prior to thrombolysis.

Although none of these studies was randomized or prospective in design, all these studies reported few, if any, complications and suggested an increased survival. In our study, in patients receiving thrombolysis, the rate of major bleeding complications was significantly higher than in patients who did not receive thrombolysis. We believe that the reason for this is that all our patients had been subjected to chest compressions during resuscitation, whilst in other studies, some patients had also been enclosed who had only received defibrillation.

Two recently published studies demonstrate the safety of thrombolytic treatment during ongoing CPR. In a prospective study, Böttiger et al. [13] administered 50 or 100 mg rt-PA as systemic thrombolysis during out-of-hospital CPR in 40 patients with suspected AMI or pulmonary embolism. In these patients, return of spontaneous circulation (ROSC) was achieved in 68%, as compared with 44% in the controls (P < 0.05), and 15% of thrombolysed patients were discharged alive from the hospital as compared with 8% in the control group. There were no bleeding complications related to CPR. In a retrospective study, Lederer et al. [16] showed that significantly more thrombolysed patients achieved ROSC during ongoing CPR (70% vs. 51%, P = 0.001), 25% of thrombolysed patients were discharged alive from the hospital as compared with 15% in the control group. Whilst serious complications such as intracerebral or subarachnoidal haemorrhage, ruptured aortic aneurysm, pericardiac tamponade and haemothorax were found in 13% of 45 patients with thrombolysis, on autopsy, in the control group serious complications were found in 15% of cases. In another retrospective study, Voipio et al. [29] reported on 68 patients who received thrombolyic treatment in an out-of-hospital setting after cardiac arrest and CPR as a result of presumed AMI. In 64 (96%) of these cases, thrombolytic treatment was considered later to be indicated, 36 (53%) of 68 patients survived to discharge. In five (7%) patients, major bleeding complications were observed, one of them had fatal cerebral bleeding. Ruiz-Bailén et al. [30] report in a retrospective cohort study on 303 cases with AMI admitted after resuscitation, of which 67 (22%) patients received thrombolytic treatment. These patients had a significantly better outcome than patients in the group without thrombolysis (82% vs. 54%; P < 0.00001) There was no significant difference in the haemorrhagic and CPR-induced complications between the two groups.

In our study, survival over 6 months was only slightly better in patients receiving thrombolysis compared with patients who did not receive thrombolysis. Even though we tried to adjust for baseline imbalances there might be residual confounding which might even explain the observed trend. Only randomized controlled trials could prove that thrombolytic treatment in patients with CPR following cardiac arrest during AMI improves survival beyond the known effect of thrombolytic treatment. There is no doubt that thrombolytic treatment is indicated in patients with AMI who also have cardiac arrest. We believe that in carefully selected patients the prognosis can be improved by timely specific treatment.

In general, it is probable that prolonged CPR increases the chance of resuscitation trauma. In the stratified analysis, we found no clear association between bleeding complications and the duration of resuscitation. It is possible that this is due to the fact that these patients were carefully selected. We prefer to believe that the absence of this dose–response effect in patients with and without thrombolysis is most likely genuine, or that an effect at best may be very weak.

Acute angiography and angioplasty are available as alternative therapeutic intervention. Spaulding et al. [31] have shown that immediate coronary angiography with angioplasty in survivors of out-of-hospital cardiac arrest caused by myocardial infarction is safe and an independent predictor of survival.


Although our study is retrospective in design, data were collected prospectively according to a protocol using Utstein Style, so we have very few missing values which is essential to minimize bias. Prior studies were all based on retrospective chart reviews and described small patient samples.

Only a small fraction of patients with AMI who develop cardiac arrest reach the hospital after successful resuscitation. So the absolute number of patients in our large registry with this condition is relatively small, which reduces the precision of our estimates and increases the possibility of a type II error, that is not detecting an effect even though it exists. Even though the effect, both in univariate and multivariate analysis, failed short of being statistically significant at a 5% level, it is likely a true effect when looking at the 95% confidence intervals.


In patients with out-of-hospital cardiac arrest caused by acute myocardial infarction, thrombolytic therapy should not be withheld after CPR. As most complications are not life-threatening and can easily be managed, the concern of excessive bleeding should be weighed against the potential benefit of specific treatment. Patients resuscitated from cardiac arrest often present with bleeding complications and may deteriorate when thrombolytic treatment is administered. Duration of CPR does not seem to be associated with haemorrhage in patients after thrombolysis.

Received 4 February 2002; revision received 24 October 2002; accepted 30 October 2002.

 Dr Fritz Sterz, Universitätsklinik für Notfallmedizin, Allgemeines Krankenhaus der Stadt Wien, Währinger Gürtel 18-20/6D, 1090 Wien, Austria (fax : +43 1 40400 1965; e-mail: