Circumstances and Mode of In-Hospital Death following 9,914 Consecutive Patients undergoing Percutaneous Coronary Interventions

ABSTRACT: Aim. To better describe the epidemiological causes of in-hospital death after percutaneous coronary intervention (PCI) in the present stent era. Methods. Systematic review of all in-hospital deaths following PCI in North West England from 2001 to 2003. Sixty-two in-hospital deaths (0.6%) were identified from 9,914 consecutive PCIs performed during the study period. The medical records of 4 patients were missing, leaving 58 patients to be reviewed with a standard data extraction tool to determine a circumstance and a mode of death. Medical records were reviewed at each center and cases were discussed at regional consensus meetings. All the collected data were validated by random cross-checking of data by exchange site visits. Multivariate logistic regression was used to identify risk factors for deaths related to procedural complications. Results. Low output failure was the most common mode of death, occurring in 42 patients (72.4%). The circumstance of death was a procedural complication in 35 patients (60.3%), and preexisting acute cardiac disease in 23 patients (39.7%). Significant predictors of death from procedural complications were treatment of left main stem (odds ratio [OR] 13.8; p < 0.001) or graft lesions (OR 5.6; p < 0.001), and female sex (OR 3.0; p = 0.002). Conclusions. Procedural complications account for over half of all post-PCI deaths. We have identified several risk factors that may help reduce the number of deaths related to procedural complications.

J INVASIVE CARDIOL 2008;20:386–390

Key Words: in-hospital mortality; percutaneous coronary interventions; mode of death; complications
In the 30 years since its use was first reported by Grüntzig,1 percutaneous coronary intervention (PCI) has become a frequently performed therapeutic procedure in patients with coronary artery disease, with an estimated 62,780 PCI procedures performed in the United Kingdom in 2004.2 The case-mix is becoming more complex and there is also a shift toward an increasing PCI use for the treatment of coronary artery disease. Yet, the anticipated benefits of PCI must be balanced against its risk.
The mortality rate following PCI was reported at 0.56% for the U.K. during 2004.2 This has remained relatively static since 1991, fluctuating between 0.5% and 0.9% over the subsequent years. Although some deaths occur in high-risk situations (e.g., PCI after shock), others are unexpected events and occur despite apparently successful procedures. Likewise, the cause of some deaths is instantly recognizable (e.g., in-laboratory abrupt closure), whereas for others, particularly those in patients who die outside of the laboratory, the proximate cause of death is less clear. Patient counseling, optimal clinical decision-making and quality improvement activities require an accurate assessment of these risks in contemporary PCI practice.
There is scant literature about how people actually die after PCI. Some of the studies available are either small or focus on the issue of unsupported or primary angioplasty.3–5 A more recently published study by Malenka and colleagues from the Northern New England Cardiovascular Disease Study Group found that procedural complications account for half of all PCI deaths and are a particular problem for women.6 As this study involved 12,232 consecutive patients undergoing PCI, excluding only patients who underwent directional atherectomy, its findings were invaluable. However, the major limitation of the studies conducted thus far evaluating circumstances and mode of death is that much of the data emanate from an era of relatively nonaggressive use of PCI, as these studies pre-date the use of stents, platelet inhibitors and other recent advances in coronary intervention.
Therefore, to better describe how and why people actually die after PCI, in a contemporary series, we undertook a review of all in-hospital deaths after PCI in North West England from 2001 to 2003.

Methods
Patient population and data. The North West Quality Improvement Programme in Cardiac Interventions (NWQIP) is a regional consortium involving four centers (Blackpool Victoria Hospital, Blackpool; The Cardiothoracic Centre, Liverpool; Manchester Royal Infirmary, Manchester; South Manchester University Hospital, Manchester) performing adult cardiac surgery and PCIs in the north west of England. The aim of this group is to continuously improve the quality of care for patients receiving cardiac interventions using a regionally based systems approach.
Data were collected on a total of 9,914 consecutive patients undergoing PCI between August 1, 2001 and  December 31, 2003 in the North West of England. Information was collected as previously described.7 In brief, each intervention had a dataset collected which included demographics, heart disease severity, acuity, comorbidity, procedural details and outcome. Validation of data was conducted at each center, which involved checking each record for completeness and flagging back to the relevant cardiology team any erroneous data. All records entered onto the databases were also cross-checked against finance activity lists and catheterization laboratory books to ensure capture of all cases. Data were collected in each hospital and returned to a central source for analysis every 6 months. Data would be returned to the providing hospital if data completeness did not achieve a rate of 98% or above. Any missing risk factor data after acceptance into the central registry were treated as absent, and this occurred in less than 2%.
Review of in-hospital deaths. The medical records of patients who died in-hospital during the study period were reviewed by one doctor with interventional interest at each institution. None of the deaths were reviewed by the individual responsible for the case. A standardized proforma was used to determine the chronologic sequence of events leading to the patient’s death and to determine the “circumstance” and primary “mode” of death. All the cases were discussed at a regional consensus meeting every 3 months and data were validated by random cross-checking and exchange site visits using the same standardized proforma.
Circumstances of death. The circumstance of death was classified as one of the following five occurrences:
• A complication of the procedure. A complication was defined as worsened congestive heart failure, a new myocardial infarction (defined as new pathological Q-wave with CK-MB > 2 times), emergency coronary artery bypass graft surgery (CABG), new arrhythmias, vascular complications, worsened renal failure, infection or a new cerebral event at the time of, or after, the PCI leading to or associated with death.
• Preexisting acute cardiac disease. New myocardial infarction, cardiogenic shock, or a refractory arrhythmia was an indication for the procedure, with subsequent death caused by the underlying pathologic conditions when the PCI did not result in complications or clearly worsen the patient’s clinical course.
• Progression of chronic cardiac disease. Progression was defined as ongoing clinical deterioration from underlying, preexisting, chronic, progressive cardiac disease and when the PCI did not result in complications or in any way contribute to deterioration of the preexisting condition. Death resulted from the natural progression of the underlying disease.
• Acute or chronic noncardiac conditions. PCI did not result in complications or in any way contribute to deterioration of the preexisting condition, and death resulted from the natural progression of the underlying disease.
• Unable to determine. For any death not classified as one of the above categories.
Mode of death. The mode of death was classified as low output failure, ventricular arrhythmia, primary respiratory failure, bleeding, infection, neurologic event, pulmonary embolism or others. The chronologic sequence of events leading up to the death was abstracted from the case notes and entered into a proforma that was used to determine which of these possibilities was the cause of death.
Statistical analysis. Continuous data are shown as median values with 25th and 75th percentiles and comparisons were made with the Wilcoxon rank sum test. Categorical variables are shown as a percentage and comparisons were made with chi-square tests as appropriate. A multivariate logistic regression analysis was performed on the entire dataset of 9,914 procedures, using the forward stepwise technique to identify independent risk factors for death from a procedural complication.8 Candidate variables were entered into the model with a p-value < 0.1. In all cases, a p-value < 0.05 was considered significant. All statistical analyses were performed using SAS for Windows, Version 8.2.

Results
In this contemporary series of 9,914 PCI patients, stent use was high, at 93%. The use of glycoprotein inhibitors was also high, at 61.6%. In-hospital death occurred in 62 cases, for a mortality rate of 0.62%.
The medical records of 4 patients were missing, and the analysis therefore is based on 58 patients. Low output failure was the most common mode of death, occurring in 42 patients (72.4%). Other modes of death included ventricular arrhythmia (13.8%, n = 8), renal failure (5.2%, n = 3), stroke (1.7%, n = 1), respiratory failure (1.7%, n = 1), ventricular rupture (1.7%, n = 1) and others (3.5%, n = 2).
The circumstance of death was determined to be a complication of the procedure in 35 patients (60.3%), while death from preexisting acute cardiac disease was the circumstance in 23 patients (39.7%). No patient’s circumstance of death was identified as progression of chronic cardiac disease or acute or chronic noncardiac conditions. The risk of dying from a procedural complication in our series was 0.35% (35/9914). The breakdown of procedural complications that led to death is detailed in Figure 1. The most frequent complications of the procedure were stent thrombosis (n = 10) and development of a new myocardial infarction (n = 10). All except 1 patient with stent thrombosis were on dual antiplatelet agents.
Regardless of the circumstance of death, low output failure occurred with comparable frequency (68.6% of deaths from a procedural complication versus 78.3% of deaths due to preexisting acute cardiac disease, p = 0.42). Most deaths (n = 45) occurred after the patient had left the catheterization laboratory, and the majority of these deaths were from low output failure (n = 30). Deaths that occurred in the catheterization laboratory were almost exclusively from low output failure (n = 12), except for 1 death due to ventricular rupture.
The time from PCI to patient death ranged from 1 to 30 days, with a mean of 5 days. There was no difference between the patients dying as result of procedural complication and death in other circumstances. Moreover, the patients dying as a result of renal failure died more than 2 weeks after PCI on hemodialysis and had normal preprocedure renal function.
To determine how patients who died of procedural complications differed from those who died from preexisting acute cardiac disease, we examined patient and procedural characteristics (Table 1) in both groups. Patients dying because of a procedural complication had a history of prior coronary artery bypass graft (CABG) surgery (17.1% vs. 0%; p = 0.04) and underwent PCI of graft lesions (14.3% vs. 0%; p = 0.06), while patients dying from preexisting acute cardiac disease had no prior CABG surgery. Procedure-related deaths were also more common in patients with unstable angina (45.7% vs. 17.4%; p = 0.03). They were less likely to have an acute myocardial infarction (31.4% vs. 73.9%; p = 0.002) or present as a nonelective case (80% vs. 100%; p = 0.02).
In our series of 9,914 patients, 29.1% patients were female, 1.1% underwent left main stem (LMS) PCI, 3.7% underwent graft PCI, 22.8% had multivessel PCI and 11.3% underwent PCI for a chronic total occlusion (CTO). Indication for PCI was stable angina in 52.8%, unstable angina in 36.3%, acute myocardial infarction in 10.3%, and cardiogenic shock in 0.7% of the patients.
In our series of 9,914 patients, risk factors for dying as a result of a procedural complication (Table 2) were female gender, LMS lesion PCI and graft lesion PCI.

Discussion
In our series of patients, we found that just over half of the deaths that occur in the setting of PCI can be attributed to a complication of the procedure. The other deaths were a consequence of the fact that the patients were critically ill going into the procedure due to preexisting acute cardiac disease. Low cardiac output is the most frequent mode of death, irrespective of the circumstance of death, and almost exclusively explained all deaths in the catheterization laboratory. Patients at increased risk of dying from a procedure-related complication were female, and underwent PCI of the left main stem or graft lesions.
Our data show in-hospital mortality following PCI across the north west of England similar to that reported across the U.K., with a mortality rate of 0.62%. BCIS audit data from hospitals across the U.K. for the year 2004 reported an in-hospital mortality rate of 0.56% (314 deaths in 56,027 PCI cases with mortality data reported).2 If we assume that the same proportion of deaths, around 60%, reported in the U.K. are also due to complications of the procedure, then this would equate to 188 deaths resulting from complications related to the procedure, which may have been preventable. 
The majority of deaths attributed to complications of procedure are a result of acute stent thrombosis and periprocedural myocardial infarction. Acute stent thrombosis resulting in in-hospital death (10/9,914 = 0.001%) is low and could be avoided by optimizing stent deployment and pharmacotherapy. Periprocedural myocardial infarctions are due to undocumented stent thrombosis, side branch occlusion and the slow reflow phenomenon following PCI and could also be avoided to some extent by technical advances in PCI and pharmacotherapy.
Malenka and colleagues6 described the mechanisms and circumstances of death in 121 patients who died after 12,919 PCIs performed in Northern New England from 1989 to 1993. Approximately one-half (53.7%) of the deaths were procedure-related, and the major mode of death was low output failure (66.1%). These results are similar to our own and suggest that even with improvements in PCI techniques and pharmacological advances, little has changed with respect to the circumstances and mode of death following PCI. Again, as with our findings, the work from Northern New England found that most deaths occurred after the patients had left the catheterization laboratory, and any deaths that did occur in the catheterization laboratory were almost exclusively associated with low output failure.
The main difference between our work and that of Malenka and colleagues involves the risk factors identified as predictors of death from a complication. Increasing age, female gender, multivessel disease, emergency procedure and intervention of the left anterior descending artery were found to be independent predictors of procedure-related death in Malenka's paper. However, of these risk factors, only female gender was identified as a predictor in our series, while we identified PCI of the LMS and graft lesions as significant risk factors. These differences could partly be due to sample size, with Malenka’s study involving 65 patients who died due to a complication of the procedure, while we only had 35 patients. However, two big changes in clinical practice could also explain why these risk factors are different. Firstly, the recent widespread use of stents has produced improved outcomes for many patients undergoing PCI and thus changed the characteristics of patients at risk of dying from a complication of the procedure, whereas none of the patients in Malenka’s study received a stent. Secondly, contemporary PCI, albeit in small numbers, is increasingly being performed on LMS or graft lesions and as our results indicate, this is a challenging area that carries an increased risk of death for patients. Previous work by our group has shown that PCI of  the LMS or graft lesions are associated with an increased risk of major adverse cardiac events.7 Recent technological advances (use of distal protection devices, novel stent deployment techniques and use of intra-aortic balloon pumps) may help to reduce some of the deaths seen in PCI of LMS and graft lesions. Female gender still remains an independent predictor of in-hospital mortality following PCI in the stent era, possibly due to the fact that female patients have higher-risk profile (advanced age, multiple risk factors and smaller vessel size) going in for PCI. These may be reduced to a certain extent by modifying the risk profile and early identification and treatment of coronary artery disease.
Abrupt closure has been historically a significant cause of death following PCI. Kahn and associates3 reported on a series of 30 deaths from 1988 to 1991 in 5,000 PCIs at a single institution. Patients with prior CABG, post myocardial infarction angina, and cardiogenic shock were not included in their series. Abrupt vessel closure was thought to be the mechanism of death for 67% of these patients, and 93% of deaths were reported as procedure-related. Ellis and coworkers4 reported 32 deaths from a tertiary care database of 8,032 patients having undergone PCI, excluding primary angioplasty. Eighty-one percent of these patients had acute vessel closure resulting in cardiac death.
Malenka and colleagues speculated that with the introduction of stents, the incidence of acute vessel closures leading to death would decrease. This would certainly appear to be the case with only 1 death in our series documented as a complication of acute vessel closure related to the procedure. This equates to just 2.9% of all the deaths related to a complication of the procedure being explained by acute vessel closure. Previous work done at one of the four institutions involved in this study between 1993 and 2000 found that nearly half (46%) of the deaths related to a complication of the procedure were due to acute vessel closure.9 The previous work had a much lower stent use rate of 48%, and would add further evidence to suggest that the risk of acute vessel closure has been reduced with the use of stent implantation.
Study limitations. There are a few limitations to our study. First, we relied on a retrospective assessment of the medical records of deceased patients to determine mode and circumstance of death, and establishing mode and circumstance of death is sometimes difficult, even with autopsy notes. Also, the data collected were subjected to local validation, however, they were not subjected to external validation and could be susceptible to bias in the determination of the circumstance of death. Being a retrospective study, accurate risk stratification prior to PCI based on angiographic features is difficult. Another limitation is the fact that we only assessed in-hospital deaths, and it is possible that additional procedure-related deaths occurred shortly after discharge. A final limitation is that with only 35 deaths recorded that were related to a complication of the procedure, this may not be sufficient for an accurate prediction.10

Conclusion
In conclusion, we have reported the circumstances and mode of death following a large group of contemporary PCI patients and identified the risk factors associated with the complication of the procedure. In-hospital mortality following PCI remains constant despite an increasingly high-risk case mix and advances in PCI techniques. However, about half of all deaths are still related to procedural complications, which may be preventable, despite the high rates of stent usage and antiplatelet therapy. Further work is needed to focus on improvements in PCI of the LMS and graft lesions, which may help lead to reductions in procedure-related deaths in these patients.
Acknowledgments. We would like to acknowledge the cooperation given to us by all the Consultant Interventional Cardiologists in the region. Blackpool Victoria Hospital: Dr. M. Brack, Dr. A. Chauhan, Dr. G. Goode, Dr. D. Roberts. The Cardiothoracic Centre-Liverpool: Dr. R.G. Charles, Dr. D.T. Connelly, Dr. M. Fisher, Dr. J.L. Morris, Dr. W.L. Morrison, Dr. N.D. Palmer, Dr. R.A. Perry, Dr. D.R. Ramsdale, Dr. R.H. Stables. Manchester Royal Infirmary: Dr. B. Clarke, Dr. N. Curzen, Dr. R. Khattar, Dr. L. Neyses, Dr. F. Ordoubadi. South Manchester University Hospital: Dr. D. Bennett, Dr. N. Brooks, Dr. H. Lee, Dr. R. Levy, Dr. B. Prendergast, Dr. S. Ray, Dr. D. Ward.
We would also like to thank, for their considerable efforts, Sue Arthur, Chantelle Bailey, Andrew Beaumont, Suzanne Chaisty, and Veronica Windsor who maintain the quality and ensure completeness of data collected in our Percutaneous Coronary Intervention Registry.