Article Text

Download PDFPDF

Original article
Determinants of excess mortality following unprotected left main stem percutaneous coronary intervention
  1. O A Alabas1,
  2. R A Brogan1,2,
  3. M Hall1,
  4. S Almudarra1,
  5. M J Rutherford3,
  6. T B Dondo1,
  7. R Feltbower1,
  8. N Curzen4,
  9. M de Belder5,
  10. P Ludman6,
  11. C P Gale1,2
  12. on behalf of the National Institute for Cardiovascular Outcomes Research7
  1. 1MRC Bioinformatics Unit, Leeds Institute of Cardiovascular and Metabolic Medicine (LICAMM), University of Leeds, Leeds, UK
  2. 2Department of Cardiology, York Teaching Hospital NHS Foundation Trust, York, UK
  3. 3Department of Health Sciences, University of Leicester, Leicester, UK
  4. 4Department of Cardiology, University Hospital Southampton NHS FT & Faculty of Medicine, University of Southampton, Southampton, UK
  5. 5Department of Cardiology, South Tees Hospitals NHS Foundation Trust, UK
  6. 6Department of Cardiology Queen Elizabeth Hospital, Birmingham, UK
  7. 7National Institute for Cardiovascular Outcomes Research (NICOR), University College, London, UK
  1. Correspondence to Dr Oras Alabas, MRC Bioinformatics Unit, Faculty of Medicine and Health, University of Leeds, Leeds LS2 9JT, UK; o.alabas{at}leeds.ac.uk

Abstract

Objective For percutaneous coronary intervention (PCI) to the unprotected left main stem (UPLMS), there are limited long-term outcome data. We evaluated 5-year survival for UPLMS PCI cases taking into account background population mortality.

Methods A population-based registry of 10 682 cases of chronic stable angina (CSA), non-ST-segment elevation acute coronary syndrome (NSTEACS), ST-segment elevation myocardial infarction with (STEMI+CS) and without cardiogenic shock (STEMI−CS) who received UPLMS PCI from 2005 to 2014 were matched by age, sex, year of procedure and country to death data for the UK populace of 56.6 million people. Relative survival and excess mortality were estimated.

Results Over 26 105 person-years follow-up, crude 5-year relative survival was 93.8% for CSA, 73.1% for NSTEACS, 77.5% for STEMI−CS and 28.5% for STEMI+CS. The strongest predictor of excess mortality among CSA was renal failure (EMRR 6.73, 95% CI 4.06 to 11.15), and for NSTEACS and STEMI−CS was preprocedural ventilation (6.25, 5.05 to 7.75 and 6.92, 4.25 to 11.26, respectively). For STEMI+CS, the strongest predictor of excess mortality was preprocedural thrombolysis in myocardial infarction (TIMI) 0 flow (2.78, 1.87 to 4.13), whereas multivessel PCI was associated with improved survival (0.74, 0.61 to 0.90).

Conclusions Long-term survival following UPLMS PCI for CSA was high, approached that of the background populace and was significantly predicted by co-morbidity. For NSTEACS and STEMI−CS, the requirement for preprocedural ventilation was the strongest determinant of excess mortality. By contrast, among STEMI+CS, in whom survival was poor, the strongest determinant was preprocedural TIMI flow. Future cardiovascular cohort studies of long-term mortality should consider the impact of non-cardiovascular deaths.

View Full Text

Statistics from Altmetric.com

Introduction

Improved stent technology, the de novo presentation of unprotected left main stem (UPLMS) coronary disease at primary percutaneous coronary intervention (PCI) and evidence from randomised trials reporting good outcomes among higher risk patients has increased the number of patients who receive PCI to the UPLMS.1–3 Among emergent, urgent and elective cases of UPLMS PCI, procedural success is high with evidence of over 95% technical success and excellent short-term outcomes.4

However, there is a paucity of representative data regarding the longer-term outcomes following UPLMS PCI. In part, this is due to the inherent bias of small observational cohorts and difficulty in generalising results from the highly selected cohorts recruited into randomised trials and because long-term survival studies of UPLMS PCI typically report all-cause mortality.5 ,6 The latter point is of particular importance when, nowadays, the dominant cause of death after PCI is non-cardiovascular and if not accounted for, the efficacy of UPLMS PCI may be underestimated.

While cause-specific mortality records can help ascertain the effect of an intervention on cardiovascular outcomes, this approach has its limitations. Cause-specific mortality records may be difficult to ascertain and when available are limited to trials or if obtained from administrative data can be biased by misclassification of the cause of death.7 An alternative method to estimate cause-specific outcomes is relative survival, which adjusts for the expected rates of death in the general population. Using a relative survival approach and all cases of UPLMS PCI within the UK healthcare system, we aimed to report the rates of relative survival and then quantify the determinants of excess mortality among emergent, urgent and elective cases of UPLMS PCI. To achieve this, we matched cases of UPLMS PCI from the British Cardiovascular Intervention Society (BCIS) national registry of PCI to life table data according to age, sex, year of procedure and country for the UK general populace of 56.6 million people. This allowed mortality and factors specifically associated with UPLMS disease and its treatment to be studied.8

Methods

Patients, setting and inclusion criteria

Participation in BCIS is mandated for all operators and all National Health Service (NHS) Providers in the UK. The sampling frame consisted of all cases from 1 January 2005 to 1 July 2014.9 Data for every PCI performed, comprising 113 core fields9 were collected prospectively, encrypted and transferred online to a central database at the National Institute for Cardiovascular Outcomes Research (NICOR). Patients included had PCI to a diseased UPLMS and were aged 18–100 years. Cases of UPLMS PCI were those in whom the LMS was the target vessel and who did not have a patent graft to any left-sided coronary artery (figure 1).9 For those with multiple records, the first admission was used. According to the joint European Society of Cardiology/American College of Cardiolog (ACC) consensus statement guidelines for definition for myocardial infarction, we grouped cases as chronic stable angina (CSA), non-ST-elevation acute coronary syndrome (NSTEACS) and ST-elevation myocardial infarction (STEMI).10 To minimise bias due to the inclusion of patients with cardiogenic shock in the STEMI group, we subdivided STEMI cases into those with (STEMI+CS) and without cardiogenic shock (STEMI−CS); both groups only included patients who received primary PCI. The diagnosis of cardiogenic shock was clinical and included a systolic blood pressure <100 mm Hg, pulse >100 bpm, in a patient who was cool and clammy or requiring inotropes, intra-aortic balloon pump or other cardiopulmonary support.

Figure 1

Consort diagram of cohort derivation. PCI, percutaneous coronary; LMS, left main stem; UPLMS, unprotected left main stem.

Mortality and follow-up

All-cause mortality data for UPLMS PCI cases were extracted through linkage to the Office for National Statistics using each patient's unique pseudonymised NHS number. Patients were followed-up for their vital status up to 5 years after PCI, with censoring at the end of follow-up on 1 July 2014 (see online supplementary table 1a, Appendix). Survival time was defined as the duration between the date of the procedure and the date of death or censoring.

Relative survival

Relative survival was defined as the observed survival among cases of UPLMS PCI divided by the expected survival of the comparable UK populace and expressed as relative survival rates (RSRs).11 A RSR of 100% implies that cases of UPLMS PCI have survival rates equal to that of the matched disease-free background population. Observed survival was estimated using the actuarial method which calculates the observed survival in time intervals from the effective number of patients at risk in that particular interval and the expected survival by the Ederer II method.11 For expected survival, country-specific population mortality rates of the UK were based on life tables from the Office for National Statistics matched to the cohort by single year of age, sex and year of procedure.

Excess mortality

Excess mortality provides a measure of the additional hazard associated with a procedure or treatment and is excess mortality rate ratio (EMRR). Evidence of excess mortality is observed when the EMRR is >1. For example, an EMRR of 1.5 for men/women indicates that men experience 50% higher excess mortality than women. The statistical model comprised generalised linear regression models, collapsed (life table) data and a Poisson error structure.11 First, we fitted a baseline model comprising age, sex and year of procedure. Each of the following covariates were then separately fitted into the baseline model: previous myocardial infarction, previous PCI, diabetes mellitus, left ventricular systolic (LVSD) function, number of vessels attempted, number of stents, renal failure (defined as serum creatinine >200 µmol/L), cardiogenic shock preprocedure and postprocedure, degree of LMS stenosis preprocedure and postprocedure (the presence of stenosis was assigned if a vessel scored >50% on the effective stenosis), thrombolysis in myocardial infarction (TIMI) flow in the infarct-related artery preprocedure and postprocedure, preprocedural ventilation, use of GPIIb/IIIa inhibitors, use of intravascular ultrasound (IVUS) and fractional flow reserve (FFR). The proportional hazards assumption was tested by including interaction terms between the three baseline variables (age, sex and calendar year) and follow-up time, and using the likelihood ratio test; there were no time dependent effects (p>0.05) therefore no interaction terms were added. To mitigate bias due to missing data, we generated 20 multiply imputed data sets by chained equations.12 All tests were two-tailed, the level of statistical significance prespecified at 5% (p<0.05) and estimates derived with 95% CI, statistical analyses were performed using Stata V.13.1 (StataCorp).

Ethics

Ethical approval was not required under NHS research governance arrangements. NICOR which includes the BCIS database (Ref: NIGB: ECC 1-06 (d)/2011) has support under section 251 of the NHS Act 2006 to use patient information for medical research without consent.

Results

Among 10 682 cases of UPLMS PCI across 89 providers there were 3799 (35.5%) CSA, 5114 (47.8%) NSTEACS, 1020 STEMI−CS and 749 STEMI+CS of whom 69.1% were male. Mean respective ages (SD) were 69.3 (11.2), 72.4 (12.1%), 68.0 (13.7) and 68.2 (12.7) years (table 1). Over 26 105 person-years and median follow-up of 2 years, 2872 (25.9%) died. The crude 5-year relative survival was 93.8% for CSA, 73.1% for NSTEACS, 77.5% for STEMI−CS and 28.5% for STEMI+CS (figure 2). The number of cases in England, North of Ireland, Scotland and Wales was 9736 (87.1%), 566 (5.2%), 439 (4.1%) and 387 (3.6%), respectively.

Table 1

Baseline characteristics for cases of UPLMS PCI by CSA, NSTEACS, STEMI−CS and STEMI+CS

Figure 2

Unadjusted 5-year cumulative relative survival stratified by chronic stable angina (CSA), non-ST-segment elevation acute coronary syndrome (NSTEACS), ST-segment elevation myocardial infarction without cardiogenic shock (STEMI−CS) and STEMI with cardiogenic shock (STEMI+CS), by predetermined time points.

Five-year relative survival by age, sex and year

Five-year relative survival was worse among the elderly (figure 3). For cases of CSA aged <55 and >75 years it was 96.3% vs 96.2% and for NSTEACS 84.1% vs 71.2%, for STEMI−CS 90.1% vs 74.8% and for STEMI+CS 41.2% vs 25.0% (figure 3). For females and males, survival was 96.2% vs 92.8% for CSA, 70.2% vs 74.5% for NSTEACS, 60.0% vs 55.8% for STEMI−CS and 36.4% vs 25.8% for STEMI+CS (figure 4). Between 2005/2006 and 2009/2010, survival rates improved; increasing for CSA (90.8%–95.5%), NSTEACS (76.5%–72.5%), STEMI−CS (72.2%–76.4%) and with the greatest improvements among STEMI+CS (23.9%–32.4%) (figure 5).

Figure 3

Unadjusted 5-year cumulative relative survival stratified by (A): chronic stable angina (CSA), (B) non-ST-segment elevation acute coronary syndrome (NSTEACS), (C) ST-segment elevation myocardial infarction without cardiogenic shock (STEMI−CS) and (D) STEMI with cardiogenic shock (STEMI+CS), by age.

Figure 4

Unadjusted 5-year cumulative relative survival stratified by (A) chronic stable angina (CSA), (B) non-ST-segment elevation acute coronary syndrome (NSTEACS), (C) ST-segment elevation myocardial infarction without cardiogenic shock (STEMI−CS) and (D) STEMI with cardiogenic shock (STEMI+CS), by sex.

Figure 5

Unadjusted 5-year cumulative relative survival stratified by (A) chronic stable angina (CSA), (B) non-ST-segment elevation acute coronary syndrome (NSTEACS), (C) ST-segment elevation myocardial infarction without cardiogenic shock (STEMI−CS) and (D) STEMI with cardiogenic shock (STEMI+CS), by calendar year.

Excess mortality by age, sex and year

For CSA, there was no evidence of excess mortality by age, sex or year of diagnosis. There was, however, significant excess mortality with increasing age (>75 compared with <55 years) for NSTEACS (EMRR 2.61, 95% CI 1.91 to 3.57) and STEMI−CS (3.49, 1.99 to 6.10), but not by sex and year of procedure. Among STEMI+CS excess mortality occurred with increasing age (1.73, 1.29 to 2.33), but not sex and there was a significant reduction in excess mortality for 2011/2012 (0.54, 0.33 to 0.91) and 2013/2014 (0.55, 0.33 to 0.91) compared with 2005/2006 (see online supplementary table 3a, Appendix).

Determinants of excess mortality

For CSA, excess mortality was associated with previous myocardial infarction (2.73, 1.77 to 4.21), diabetes (2.56, 1.64 to 3.97), moderate (2.43, 1.38 to 4.29) and poor left ventricular systolic function (3.90, 2.23 to 6.82), renal failure (6.73, 4.06 to 11.15) and preprocedural stenosis severity (EMRR 1.82, 95% CI 1.02 to 3.23) (table 2). There was a significant reduction of excess mortality associated with the use of IVUS (EMRR 0.46, 95% CI 0.28 to 0.76). Other variables such as the number of vessels attempted, the number of stents deployed and the use of a GPIIb/IIIa were not significantly associated with excess mortality.

Table 2

Factors associated with excess mortality for cases of UPLMS PCI by CSA, NSTEACS, STEMI−CS and STEMI+CS

For NSTEACS, excess mortality was associated with previous myocardial infarction (EMRR 1.55, 95% CI 1.34 to 1.80), diabetes (1.66, 1.43 to 1.95), moderate (2.34, 1.86 to 2.93) and poor left ventricular systolic function (3.65, 2.96 to 4.51), renal failure (3.25, 2.75 to 3.84), preprocedural ventilation (6.25, 5.05–7.75) and preprocedural LMS disease severity (2.09, 1.64 to 2.66). Reduced excess mortality was significantly associated with IVUS (EMRR 0.47, 95% CI 0.39 to 0.57) and FFR (0.44, 0.25 to 0.75).

Among cases of STEMI−CS, excess mortality was associated with diabetes (EMRR 1.69, 95% CI 1.15 to 2.48), renal failure (2.27, 1.20 to 4.27), TIMI 1 flow versus normal flow (1.99, 1.01 to 3.92), preprocedural ventilation (6.92, 4.25 to 11.26) and the LMS disease severity (1.60, 1.01 to 2.41). Factors which were associated with reduced excess mortality were the deployment of one (0.31, 0.18 to 0.54) or more stents (0.32, 0.20 to 0.53), the use of GPIIb/IIIa inhibitors (0.42, 0.29 to 0.59) and the use of IVUS (0.43, 0.24 to 0.74).

For STEMI+CS, excess mortality was associated with diabetes (EMRR 1.34, 95% CI1.04 to 1.72), renal failure (1.77, 1.13 to 2.75), preprocedural ventilation (2.03, 1.66 to 2.50) and any degree of reduction of preprocedural TIMI flow versus the normal flow (TIMI 0: EMRR 2.78, TIMI 1: 2.32, TIMI 2: 2.00). Factors significantly associated with reduced excess mortality were multivessel PCI (EMRR 0.74, 95% CI 0.61 to 0.90), deployment of stents versus no stent (one stent: 0.45, 0.32 to 0.62), more than one stent (0.42, 0.31 to 0.57) and IVUS (0.28, 0.17 to 0.46). Neither preprocedural degree of LMS stenosis nor GPIIb/IIIa inhibitors were associated with excess mortality.

Discussion

This is the first population-based study estimating long-term relative survival for patients who received PCI to an UPLMS. Relative survival provides an objective measure of the proportion of patients dying from direct or indirect consequences of a disease without requiring a record of the precise cause of death.11 To date, studies of UPLMS PCI have focused on observed survival and, therefore, reported outcomes include deaths related to the procedure and ‘natural’ deaths occurring in the cohort under study.2 ,13 ,14 Our study provides new insights through the analysis of nationwide prospective, consecutive series registry data accounting for populace mortality data.

Specifically, we found that survival after UPLMS PCI for elective patients with CSA was very high (over 90% survived to 5 years) and approached that of the age, sex year and country matched disease-free general population. While the presence of an acute coronary syndrome (either NSTEACS or STEMI) was associated with reduced longer-term survival, for NSTEACS and STEMI−CS survival was similar (about 75% survived to 5 years). By contrast, emergent cases of STEMI presenting with cardiogenic shock had very poor survival, which was evident immediately after PCI and persisted for many years (about 30% survived to 5 years). Data from this study provide real world evidence to substantiate the ACC/American Heart Association (AHA) guidelines, which have upgraded PCI for UPLMS in specific circumstances from a class III to a class I or IIa procedure.15

However, our study did identify an improvement among cases of STEMI+CS, and it is possible that operators and hospital services are more familiar with the urgent management of such cases. It is also possible that improved stents, deployment techniques (for complex anatomy) as well as more potent pharmacological treatments have, in part, facilitated the temporal improvements. Notwithstanding this, the ‘accelerated failure’ and poor long-term survival among STEMI+CS were very clear. It appears, therefore, that the greatest gains for improved outcomes are among high-risk cases of complex PCI. Survival for these cases was, however, constrained by co-morbidity, poor preprocedural TIMI flow in the infarct-related artery and the necessity for preprocedural ventilation, but was associated with an improvement of about 50% with multivessel PCI. Given that trials testing the efficacy of the intra-aortic balloon pump have failed to reach their primary end points and that there is insufficient evidence for the use of percutaneous assist devices, a greater focus on technologies which support the myocardium (thereby allowing optimal infarct and non-infarct-related PCI), improved stent design and enhanced operator experience is needed.16

Whereas, for elective cases of UPLMS PCI, attention to co-morbidities (previous MI, diabetes and renal failure), optimisation of left ventricular systolic function and careful evaluation of preprocedural stenosis severity (using IVUS or FFR) are likely to be key to maintaining the present rates of survival. Addressing these factors and using them to help predict a patients' clinical outcome will provide the opportunity for clinicians to discuss in greater detail the risks and benefits of the intended procedure. However, as survival rates are already very high among this group, future absolute gains are likely to be small.

For NSTEACS, factors that negatively impacted on long-term survival were prior myocardial infarction, diabetes, LMS stenosis >50%, moderate and poor left ventricular systolic function, renal failure, cardiogenic shock and the requirement for mechanical ventilation. As for elective cases, the use of IVUS and FFR was associated with a >50% reduction in excess mortality. Although our study design cannot determine a cause and effect relationship, this observation supports guideline recommendations that careful attention to the coronary anatomy and stent deployment are central to good outcomes. Even so, it is possible that in our study, more stable, lower risk, patients were more likely to receive IVUS and FFR.17 ,18

For all types of clinical presentations except STEMI+CS, we found that multivessel PCI was not associated with a survival advantage. For STEMI complicated by cardiogenic shock, multivessel PCI was associated with, on average, a lower relative excess mortality of 26%. This novel association warrants further large-scale evaluation, especially when present data have failed to eliminate the clinical uncertainty about the most appropriate way to treat patients with multivessel disease. That is, while previous studies have not confirmed the benefit of more complete revascularisation in the context of cardiogenic shock, recent trials (among those without CS) have questioned the conventional view of lesion only revascularisation in STEMI.19 ,20

Strengths and limitations

Even though relative survival and excess mortality are novel concepts for the evaluation of cardiovascular outcomes, these techniques are well established in cancer epidemiology. Relative survival is an underused tool in cardiovascular outcome reporting,8 which in an era of evidence-based practice and an ageing ‘survivorship’ population merits further attention. The use of relative survival for this study has allowed higher resolution estimation of survival and excess deaths specifically due to UPLMS disease and its percutaneous treatment without requiring potentially unreliable ‘cause of death’ data.

While this study has other strengths, including the size and quality of data, there were limitations. A high prevalence of the index disease among the general population will over-inflate survival estimates.11 Although cardiovascular disease is prevalent, this is unlikely to be the case for UPLMS disease per se—being identified at diagnostic angiography in 4%–6% of cases and estimated at 15 cases per 100 000 population/year.21 Selection bias may have been introduced through the identification and consent of patients, which may lead to a healthier cohort than expected.

Conclusions

In the largest long-term outcomes study of UPLMS PCI, and after adjustment for non-cardiovascular death, survival for patients with CSA was excellent and approached that of the general population. This contrasted with emergency cases and, in particular, STEMI+CS where, despite temporal improvements, survival was poor. For NSTEACS and STEMI−CS, preprocedural ventilation was the strongest determinant of excess mortality, in contrast to STEMI+CS where survival was poor and the strongest determinant was preprocedural TIMI flow.

Key messages

What is already known on this subject?

  • Mortality after percutaneous coronary intervention (PCI) to the unprotected left main stem (UPLMS) is higher among emergency and urgent cases than elective cases and it is especially high among patients with cardiogenic shock. Following PCI, however, the dominant cause of death is non-cardiovascular.

What might this study add?

  • After adjusting for background population mortality, we found that long-term survival following UPLMS PCI for elective cases was excellent, approached that of the background populace and was significantly predicted by co-morbidity. For non-ST-segment elevation acute coronary syndrome and ST-segment elevation myocardial infarction (STEMI) without cardiogenic shock, the requirement for preprocedural ventilation was the strongest determinant of excess mortality. For STEMI with cardiogenic shock, where survival was poor, the strongest determinant was thrombolysis in myocardial infarction flow.

How might this impact on clinical practice?

  • Greater attention to specific determinants of excess mortality, such as diabetes, renal failure and coronary anatomy, according to whether a case is emergent, urgent or elective will help improve survival following UPLMS PCI. Knowledge of clinical presentation-specific factors associated with excess mortality will allow better forecasting of outcomes for patients with UPLMS disease. The poor and persistently low survival among STEMI with cardiogenic shock requires greater clinical attention.

Acknowledgments

The extract from the British Cardiovascular Intervention Society (BCIS) registry was provided through the BCIS/National Institute for Cardiovascular Outcomes Research. We gratefully acknowledge all the hospitals in the United Kingdom for their contribution of data to BCIS.

References

View Abstract

Footnotes

  • Study design: Prospective population -based linked cohort study using data from the British Cardiovascular Intervention Society (BCIS) database, January 2005 to July 2014.

  • Contributors I can confirm that all authors have contributed to the article and have read the final version.

  • Funding MH and TBD are funded by the British Heart Foundation (Project Grant PG/13/81/30474). CPG is funded by the National Institute for Health Research (NIHR-CTF-2014-03-03) as Associate Professor and Honorary Consultant Cardiologist. The British Cardiovascular Intervention Society is commissioned by the Health Quality Improvement Partnership as part of the National Clinical Audit and Patient Outcomes Programme.

  • Competing interests NC declares unrestricted research grants from Boston Scientific, HeartFlow, Haemonetics, Medtronic, St Jude Medical, and speaker fees/consultancy from St Jude Medical, HeartFlow, Haemonetics, Lilly/D-S.

  • Ethics approval National Institute for Cardiovascular Outcomes Research.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.