Replacing warfarin for better or worse: identifying patient factors and future directions

Care models must consider patient factors to ensure safe and effective use of patient-administered anticoagulants

In Australia, atrial fibrillation (AF) affects 1%–2% of the population, placing them at a fivefold increased risk of stroke.1 The prevalence of AF increases sharply with age, with 50% of patients aged 75 years or over.2 Warfarin has been used clinically for more than 70 years and is the mainstay anticoagulant for prophylaxis of stroke as well as for the management of AF. However, in recent years, the novel oral anticoagulants (NOACs) dabigatran, rivaroxaban and apixaban have been developed. These promise more efficacy than warfarin in stroke prevention, while being safer and easier to use.3

A recent review by the Australian Government Department of Health and Ageing provided perspectives on current and future options for improving the use of anticoagulation therapies in managing patients with AF.2 In this article, we focus on strategies to improve anticoagulation management in patients with AF in light of the increasing utility of NOACs.

Warfarin in clinical practice

Despite the availability of oral alternative therapies, warfarin — a vitamin K antagonist — remains efficacious and cost-effective in reducing the incidence of stroke in patients with AF.4 However, warfarin has a narrow therapeutic window, resulting in an unpredictable anticoagulant effect that necessitates frequent monitoring and dosage adjustment.4 Further, warfarin is associated with the clinical trade-off between the risk of ischaemic stroke and an increased risk of bleeding, particularly intracranial haemorrhage. As a consequence, a range of risk factors that can influence warfarin stability have been identified and are well known in clinical practice.5,6 These complexities have resulted in underuse of warfarin, particularly among older people, where the risk of cerebrovascular and cardiovascular events is highest.7

Although warfarin has been used clinically for many decades, there is an ongoing need to improve the monitoring and management of patients who receive it. Optimisation of time in the therapeutic range (TTR) is the key to improving patient outcomes. This requires identification of factors that can influence TTR and the incorporation of management strategies.2 Although multiple comorbidities and concomitant medications are regularly cited as reasons for poor TTR, other relevant patient factors, such as cognitive impairment and inadequate health literacy, should be considered, as these have been shown to influence anticoagulation control.8

Clinical trials of the new anticoagulants

Dabigatran, a thrombin inhibitor, and rivaroxaban and apixaban, both factor Xa inhibitors, are the first oral alternatives to warfarin to be made available for the prevention of stroke and systemic embolism in patients with non-valvular AF. All three NOACs have demonstrated effectiveness in multinational randomised clinical trials although, to date, no studies have directly compared them. Pivotal trials for dabigatran (RE-LY),9 rivaroxaban (ROCKET AF)10,11 and apixaban (ARISTOTLE)12 recruited patients from 40 different countries, with the aim of evaluating their respective efficacy for the prevention of stroke and the risk of haemorrhage compared with warfarin (Box). The results look promising, with all three NOACs demonstrating non-inferiority to warfarin for stroke and systemic embolism prevention.9–12 However, their most consistent clinical benefit is that they are superior to warfarin when it comes to the rate of intracerebral haemorrhage and haemorrhagic stroke.10,12 Further, they have the notable practical advantages of involving fixed dosages and not requiring international normalised ratio monitoring.3

Although these results show clear benefits, there are some significant differences between the trials, including different definitions at end points such as major bleeding, differences in follow-up periods, and different distribution of patients in various countries, making direct comparison difficult.13,14 Most notably, consideration must be given to the quality of warfarin management and stability in the control arms, as this varies markedly and is also at variance with that seen in clinical practice. The mean TTRs for patients taking warfarin in the three NOAC trials9–11 were 64%, 55% and 62%, respectively. These proportions are significantly lower than that found in the Australian population, which has been reported to be about 70% and above.2,15 Across these studies, when TTR was in the higher quartile, patients had an overall better level of health management. This has implications for the generalisability of these findings to an Australian population and, consequently, to the cost-effectiveness in an Australian setting.2

Challenges for chronic AF anticoagulation management in the community

Whatever the choice of anticoagulant, it is imperative that patient-related factors are considered in the selection and management of chronic AF therapies that are primarily self-administered in the community. This is particularly relevant when considering translation from clinical trial settings into the wider clinical practice environment. It is known that clinical use will be different from that within the controlled clinical trial setting. Recently published data on local experience in the use of dabigatran has shown this to be the case.16 Compared with the participants in the RE-LY trial,9 the Australian cohort had older participants, fewer patients with significant renal impairment, and a higher TTR of 70%. These differences will play a role in the real world, outside of the control of the clinical trial arena.

The challenges faced by Australian health care providers are that many patients with AF are older, with multiple comorbidities and on a range of other chronic therapies, and are self-managing their anticoagulation in the community.17 Selection of the appropriate therapy requires not only consideration of specific aspects of the individual drug but, even more importantly, the services and supports for the ongoing management and care after treatment initiation.

In the case of chronic disease management in older patients, approaches to improving anticoagulation must not only focus on medication initiation but also on the patients and the changes that inevitably occur in their ability to self-manage.17,18 A wide range of recommendations from the Department of Health and Ageing review aims to improve health outcomes and remove barriers to the use of warfarin by improving the quality of use.2 We believe that these recommendations should be progressed. A number of them specifically relate to strategies to improve chronic anticoagulation management, including (i) a requirement for national guidelines that are disseminated and implemented; (ii) guidelines for patient management; (iii) availability and coordination of community-based anticoagulation support services for patients for whom therapy is initiated in the community; (iv) optimisation of TTR; and (v) a nationally endorsed shared care model for warfarin monitoring and management.2

We support these recommendations and believe that increased emphasis should be given to improvements in strategies focusing specifically on patient-related factors that influence the management of oral anticoagulation when self-administered in the community. In addition to demographic and clinical risk factors, we identified the following specific patient factors as independent risk factors for instability of anticoagulation and bleeding risk in response to warfarin: cognitive impairment, social isolation, depressed mood, inadequate health literacy, and functional dependence.8 There was a 3.4-fold increased risk of bleeding when multiple psychosocial factors were present.8 Strategies to improve chronic AF anticoagulation management in the community must take patient factors into account, not only at the drug selection and initiation of treatment stage but also through ongoing assessment as the patient’s age increases.19 Cognitive decline has been linked to ageing. Patients with progressive cognitive impairment, and who remain responsible for their own medication management, may have increasing difficulty routinely administering and monitoring any of their medications, not just warfarin. Including patient factors in national guidelines and shared care models is fundamental to improving anticoagulation in AF through regular patient assessment.

Conclusion

Patient factors must be taken into account not only at the point of drug selection but as a routine component of care. These factors include poor cognition, depressed mood, functional dependence, social isolation and inadequate health literacy. Improved models of care are required to monitor and manage patient factors to ensure safe and efficacious anticoagulation, irrespective of the specific drug involved. Translation of clinical trial findings into the broader clinical environment must also include consideration of all the factors that influence patient variability.

Further, improved stroke prevention and AF management are required and include medication innovation as well as innovation in models of patient care. Irrespective of whether warfarin or a NOAC is chosen, the challenges for improving management and monitoring in potentially complex patient cohorts remain.

Summary of key studies comparing dabigatran, rivaroxaban or apixaban with warfarin2

Study design

Baseline characteristics

Comorbidities

Reduction of total stroke or systemic embolism, HR or RR (95% CI)

Reduction of bleeds, OR (95% CI)

RE-LY9

Prospective, multicentre, multinational, randomised, parallel-group trial: dabigatran (110 mg or 150 mg twice daily) v adjusted warfarin dose (N = 18 133)

Mean age, 71.5 years; male, 64%

History of TIA/stroke, 20%; previous MI, 17%; hypertension, 79%; diabetes, 23%

110 mg: RR, 0.91 (0.74–1.11)

110 mg: major bleed, 0.80 (0.69–0.93); major GI bleed, 1.10 (0.86–1.41); ICH, 0.31 (0.20–0.47); all-cause mortality, 0.91 (0.80–1.03)

150 mg: RR, 0.66 (0.53–0.82)

150 mg: major bleed, 0.93 (0.81–1.07); major GI bleed, 1.50 (1.19–1.89); ICH, 0.40 (0.27–0.60); all-cause mortality, 0.88 (0.77–1.00)

ROCKET AF11

Double-blinded randomised controlled trial: rivaroxaban (20 mg daily) v adjusted warfarin dose (N = 14 264)

Mean age, 73.0 years; male, 60%

History of TIA/stroke, 55%; previous MI, 17%; hypertension, 90%; diabetes, 40%

HR on PP, 0.79 (0.66–0.96); HR on ITT, 0.88 (0.74–1.03)

Major bleed, 1.04 (0.90–1.20); major GI bleed, 1.60 (1.29–1.98); ICH, 0.67 (0.47–0.93); all-cause mortality, 0.85 (0.70–1.02)

ARISTOTLE12

Double-blinded randomised controlled trial: apixaban (5 mg twice daily) v adjusted warfarin dose (N = 18 201)

Mean age, 70.0 years; male, 65%

History of TIA/stroke, 19%; previous MI, 14%; hypertension, 87%; diabetes, 25%

HR, 0.79 (0.66–0.95)

Major bleed, 0.69 (0.60–0.80); major GI bleed, 0.89 (0.70–1.15); ICH, 0.42 (0.30–0.58); all-cause mortality, 0.89 (0.80–0.99)

GI = gastrointestinal. HR = hazard ratio. ICH = intracranial haemorrhage. ITT = intention-to-treat analysis. MI = myocardial infarction. OR = odds ratio. PP = per-protocol analysis. RR = relative risk. TIA = transient ischaemic attack.

The approach to patients with possible cardiac chest pain

Chest pain is a confronting symptom for patients and clinicians alike. Some patients presenting with chest pain will have serious acute illness with a high short-term risk of mortality, but this will be excluded in most patients. Chest pain is one of the most common causes of attendance at hospital emergency departments (EDs) and a frequent cause of presentations to general practice.1 Missed diagnosis, with associated adverse outcomes, can occur when chest pain assessment is based on clinical features alone.2

Regardless of the clinical setting, a stepwise approach should be applied to patients with chest pain (Box 1). In the absence of trauma, the primary focus should be exclusion of four potentially fatal conditions: acute coronary syndrome (ACS; encompassing acute myocardial infarction and unstable angina), pulmonary embolism, aortic dissection and spontaneous pneumothorax. ACS is by far the most common of these. All these conditions may present without immediately obvious physical signs, but the latter three may be accurately excluded by rapid diagnostic testing (predominantly medical imaging). However, ACS is more challenging as it cannot be readily excluded with an acceptable level of accuracy on initial clinical evaluation or with a single investigation. After excluding these conditions, attention should be turned to chronic but serious conditions that may require additional evaluation, such as stable coronary artery disease or aortic stenosis. Next, non-life threatening conditions that may benefit from specific therapy (eg, herpes zoster, gastro-oesophageal reflux) should be considered. If the clinician is confident that all these causes have been excluded, the patient can be reassured that the chest pain is due to an insignificant cause.

Here, we focus on several evolving areas relating to the assessment of patients with possible cardiac chest pain, including risk stratification, cardiac biomarkers and the role of non-invasive testing for myocardial ischaemia and coronary artery disease. This article is aimed at all clinicians who assess patients with acute, undifferentiated chest pain. It is not a systematic review, but we have directed the reader towards these where appropriate.

The aim of chest pain assessment

There is a dichotomy in the assessment of patients with possible ACS. First, early and accurate identification of patients with ST-segment-elevation myocardial infarction (STEMI) enables provision of emergency reperfusion therapy, which has a major impact on outcome, while accurate identification of patients with other types of ACS (non-ST-segment elevation myocardial infarction [NSTEMI] or unstable angina) allows for early initiation of targeted treatment known to improve outcomes in these groups. Second, accurate exclusion of myocardial ischaemia in patients with chest pain is essential to minimise the morbidity and mortality associated with missed diagnoses, while avoiding unnecessary overinvestigation in those without the disease. However, assessment is complex because of the diversity of clinical presentations of ACS and the lack of a single diagnostic test for the entire spectrum of disease.

Despite recognition that clinical systems are imperfect, a high degree of safety in chest pain assessment processes is demanded. A recent large survey of emergency physicians suggests that the target rate of unexpected adverse outcomes in patients with a negative chest pain assessment should be < 1% at 30 days;3 this target is likely to be equally stringent in primary care. Achieving this level of safety in a timely and cost-effective fashion in an era of increasing demand on acute services presents challenges. This must be considered when the potential value and safety of new developments are assessed.

Clinical approach and risk stratification

Most current diagnostic strategies for acute chest pain focus on the identification of ACS and are based on the premise that other obvious diagnoses have been excluded with accurate clinical assessment.

Systematic reviews of the diagnostic value of clinical features in the assessment of chest pain have largely been carried out in hospital settings,4 where the prevalence of serious disease is higher than in general practice. It is widely understood that no single clinical feature or combination of features can be used to exclude ACS with sufficient sensitivity to obviate the need for further investigation. Thus, a strategic approach based on clinical risk stratification, a period of observation, electrocardiography and serial biomarker evaluation has emerged. In all settings, a 12-lead electrocardiogram (ECG) should be performed immediately in patients presenting acutely with chest pain to exclude ST-segment-elevation.

In general practice, the aim should be to differentiate patients who require urgent hospital-based assessment for possible ACS from those with more stable symptoms who may be investigated on an outpatient basis. Limited access to investigations encourages the use of clinical judgement or clinically based decision rules to triage patients who can continue to be managed safely in primary care. Several such decision rules exist, but with limited validation for use in primary care. Despite the lower prevalence of coronary disease in patients presenting with chest pain in primary care, the same limitations found in hospital-based cohorts apply to the value of clinical assessment. A recent well conducted Swedish study concluded that the accuracy of clinical assessment of chest pain by general practitioners was high, but insufficient to safely rule out coronary artery disease.5 Clinicians in general practice should refer patients promptly to hospital for assessment when features suggesting a diagnosis of ACS are present (Box 2).

The value of further investigation in general practice of patients with an acute onset or ongoing symptoms is limited, given that a normal ECG cannot exclude a significant short-term risk of an adverse outcome, and serial biomarker testing is required to exclude myocardial infarction. Nevertheless, in all settings, the resting ECG has a critical role in identifying patients with ST-segment elevation who require emergency reperfusion therapy. Patients with suspected ACS and ongoing pain, pain within the past 12 hours that has resolved but with an abnormal ECG, or other high-risk features (Box 3) should be referred to hospital as an emergency. Given the release kinetics of troponin, a single troponin test may have value in assessing patients with a normal ECG and no high-risk features who present more than 12 hours after resolution of symptoms suggestive of ACS. In such cases, appropriate mechanisms must be in place for prompt review of results and referral to hospital where necessary. If these facilities are unavailable, patients should be referred to the ED for same-day chest pain assessment.

Demographic and cardiovascular risk factors, such as age and sex, influence population risk of disease but should not unduly influence the assessment of individual patients. In the absence of a clear alternative diagnosis, most patients will require additional investigation to exclude coronary artery disease, and the critical decision is usually not whether, but with what urgency, this should be undertaken. In some countries, rapid-access chest pain assessment clinics offering early assessment of patients (usually within 14 days) have become an integral part of strategies for chest pain assessment as an alternative to ED-based assessment.6 However, these have not been widely implemented in Australia, and all acute care facilities with an ED should have an evidence-based strategic approach to assessing patients with chest pain.

Patients should be stratified as being at low, intermediate or high risk of short-term adverse outcomes in the context of possible ACS, in line with the joint guidelines of the National Heart Foundation and Cardiac Society of Australia and New Zealand (NHF/CSANZ) stratifying patients with ACS (Box 3).6 This model has performed well in the ED setting, with 30-day risks of adverse cardiac outcome of 0, 7% and 26% in these risk strata, respectively, when the criteria were strictly applied in one cohort.7 Risk stratification models may have greater utility in the ED, where the prevalence of ACS is about 10% (compared with primary care, where rates are lower) and where facilities to further assess patients at increased risk are readily available. The main limitation of this risk stratification model is that few patients qualify as low risk when the criteria are strictly applied. Alternative approaches include the Thrombolysis in Myocardial Infarction (TIMI) score, the Global Registry of Acute Coronary Events (GRACE) score and the GRACE Freedom-from-Event score.8,9 These models, derived from higher-risk populations, were not designed to identify low-risk patients who do not require detailed assessment for exclusion of ACS. Consequently, none can be relied on to identify patients who can be safely discharged from the ED without some period of observation and additional investigation. Nevertheless, risk stratification is essential to guide the appropriate use of resources based on pretest probability of ACS.

Cardiac biomarkers

Cardiac troponin levels have a central role in the diagnosis of acute myocardial infarction.10 After exclusion of ST-segment elevation and dynamic ST-segment electrocardiographic changes, serial biomarker testing identifies the remaining patients with acute myocardial infarction. Protocols for the use of serial troponin measurements have largely been based on release kinetics in experimental conditions and have tended to require waiting 6–8 hours (or longer) after presentation for the second test. Recent advances in high-sensitivity assays that allow a much shorter interval of 2 hours before the second test and incorporation of serial biomarker levels into overall risk stratification models (Box 4) have demonstrated safe accelerated processes with robust clinical outcome data.11,12 These approaches have yet to be incorporated into clinical guidelines, but almost certainly will be in the foreseeable future.

Troponin levels are considered abnormal when they exceed the 99th percentile of a healthy reference population using an assay with sufficient accuracy at this level (< 10% coefficient of variation). In practice, few available assays have possessed sufficient accuracy at this level.13 The recent development of high-sensitivity assays with this level of accuracy and lower levels of detection allows measurable troponin levels to be recorded in most of the healthy population. These assays offer the promise of being able to rule out acute myocardial infarction earlier than was possible with less sensitive assays, as well as further acceleration of risk stratification models, but with the probable cost of diminished specificity.14 This will require clinicians to have a better understanding of the causes of elevated troponin levels and the kinetics of troponin release at these new lower levels of detection, possibly by incorporating values expressing change or “delta” troponin.15 The use of delta troponin values has been incorporated into the 2011 addendum to the NHF/CSANZ guidelines, but the evidence for the best approach is still emerging.16 It is imperative that clinicians have a clear understanding of the characteristics of the local troponin assay used, as reference intervals are not transferable between different troponin assays.

Investigations for myocardial ischaemia and coronary artery disease

In two groups of patients — those who present with symptoms of ACS and in whom myocardial infarction has been excluded, and those with a stable pattern of chest pain symptoms in whom angina cannot be excluded — additional testing is required to identify those who have prognostically important coronary artery disease or unstable angina. This is an area where well established diagnostic tests exist alongside more recent developments, such as computed tomography coronary angiography (CTCA). The anatomical and pathophysiological bases for these tests are not interchangeable, with some depending on the detection of abnormal coronary blood flow (myocardial perfusion scanning) or myocardial ischaemia (stress electrocardiography and stress echocardiography), while invasive angiography and CTCA demonstrate the anatomical basis of coronary artery disease. Each investigation has different limitations depending on patient factors and the need for contrast media and ionising radiation, and the availability of each may depend on access, cost and local expertise (Box 5).

Non-invasive testing for myocardial ischaemia or coronary artery disease is of most value to patients with intermediate pretest probability of an ACS. In patients with very low risk of coronary artery disease who have symptoms of non-ischaemic pain, other causes of chest pain should be actively excluded before investigations for myocardial ischaemia or coronary atheroma are considered. Similarly, it may be futile to embark on non-invasive testing (with an attendant risk of a false negative result) in a patient with typical symptoms and a very high risk of coronary artery disease. In such cases, prompt specialist referral for consideration of an early invasive strategy should be the first step.

Investigations may identify the presence or effects of coronary artery stenosis but, where this cannot be achieved, a broader aim is to further refine risk stratification to identify patients at low risk of an adverse outcome after discharge from the hospital or ED. Exercise stress electrocardiography has become largely obsolete as a means of diagnosing reversible myocardial ischaemia, due to insufficient diagnostic accuracy, but it retains a well established role in identifying patients with chest pain who can safely be discharged from the ED.17,18 Exercise stress electrocardiography may be limited by patients’ inability to exercise at an adequate level, non-specific electrocardiographic changes (particularly in the setting of an abnormal resting ECG), and false positive results, but it remains attractive by virtue of its low cost and widespread availability.

The combination of cardiac imaging with exercise or pharmacological stress testing can increase accuracy beyond electrocardiography alone (Box 6).19–21 In the United States and Europe, cardiac magnetic resonance imaging has emerged as a safe, non-ionising and more accurate alternative to nuclear perfusion scanning, but it remains predominantly a research tool in Australia.23

CTCA is the most rapidly evolving test for assessing patients with chest pain and is the most sensitive non-invasive test for identifying coronary artery disease.22 Recent studies have shown that this technique allows patients to be safely discharged from the ED.24 A CTCA-based strategy may also be faster than other strategies, particularly when these rely on hospital admission for myocardial perfusion scanning.24,25 However, this finding is of limited value in Australia, where myocardial perfusion scanning has not been the principal investigation for chest pain assessment.

It is important to recognise some limitations of CTCA. Elevated heart rate, coronary calcium and obesity all impair image quality. The use of iodinated contrast media is risky in patients with renal impairment or in those taking metformin. In the widely cited Coronary Computed Tomographic Angiography for Systematic Triage of Acute Chest Pain Patients to Treatment (CT-STAT) trial, only 11% of the patients screened met the study’s inclusion criteria.25 Early studies suggested that CTCA should not be performed until after a second troponin measurement, as myocardial infarctions caused by moderate, rather than severe, coronary stenoses could potentially be missed.26 This emphasises that the strength of CTCA lies in excluding coronary atheroma. Furthermore, in the presence of known coronary artery disease, functional testing for ischaemia may be a more appropriate choice of investigation.27

Some centres perform CTCA with a total radiation dose of < 1 mSv, but in most centres, using general CT scanners without modern dose-reduction equipment, the total dose is likely to be significantly higher. Patients presenting to the ED, where there is an imperative to achieve a diagnostic study regardless of heart rate, may receive 10 mSv, although this is still lower than in most myocardial perfusion scans.28 There is now strong evidence that CT radiation can induce cancer.29 As CTCA could potentially be applied to more than 60% of patients presenting with chest pain in Australia, it is appropriate to remember that other tests are available and that CTCA has not yet demonstrated superiority in this setting. Nevertheless, CTCA is likely to become an increasingly important tool for ruling out significant coronary artery disease in patients with chest pain. Ongoing large clinical trials, such as the Prospective Multicenter Imaging Study for Evaluation of Chest Pain (PROMISE),30 should provide more definitive evidence in this area. Currently, Medicare regulations limit rebates to specialist referral for CTCA, and a robust system of credentialling has been introduced as a quality control measure.

Complex tests need to be appropriately incorporated into an overall strategy of risk-based chest pain assessment, integrating safe, accessible and cost-effective techniques that can accommodate the broadest range of patient presentations and comorbidities.

Conclusion

Chest pain is a common presenting symptom with many diagnostic challenges and pitfalls. Medicopolitical imperatives such as the National Emergency Access Target render the situation more complicated still. Both technology and the evidence base guiding the approach to the problem have developed considerably since the NHF/CSANZ first commissioned guidelines in this area in 2000. Clinicians can now benefit from a better understanding of risk stratification and enhanced diagnostic tools that make excluding avoidable short-term adverse events with a high degree of accuracy a realistic proposition. The challenge remains to implement these advances as widely as possible in an environment of constrained resources and increasing demand. This will be best achieved by an approach that integrates the technology and evidence into a comprehensive but straightforward and accessible strategy.

1 A pragmatic differential diagnosis of non-traumatic chest pain*

Life-threatening diagnoses that should not be missed:

Acute coronary syndrome

Acute myocardial infarction
Unstable angina pectoris

Acute pulmonary embolism
Aortic dissection
Spontaneous pneumothorax

Chronic conditions with an adverse prognosis that require further evaluation:

Angina pectoris due to stable coronary artery disease
Aortic stenosis
Aortic aneurysm
Lung cancer

Other acute conditions that may benefit from specific treatment:

Acute pericarditis
Pneumonia or pleurisy
Herpes zoster
Peptic ulcer disease
Gastro-oesophageal reflux
Acute cholecystitis

Other diagnoses:

Neuromusculoskeletal causes
Psychological causes

* This differential diagnosis is not intended to be exhaustive.

2 Case vignette

A 66-year-old man calls his general practitioner for advice. He has been treated for type 2 diabetes and primary prevention of cardiovascular disease for about 5 years. He calls from the airport where he is due to board an interstate flight but is concerned because he experienced 20 minutes of “burning” central chest discomfort while walking into the airport. He has had similar self-limiting symptoms with exertion for 4 weeks. During this time, he underwent upper gastrointestinal endoscopy that was unremarkable.

Comment: Even with the limited information available from a telephone call, and despite the atypical description of the chest pain, this patient exhibits several features suggestive of intermediate risk for an acute coronary syndrome.6 As such, he should be advised to attend hospital for assessment of chest pain without delay.

The patient is reviewed in the emergency room, where he has an unremarkable electrocardiogram and a cardiac troponin I level of 0.01 μg/L on admission, and 0.02 μg/L 6 hours and 25 minutes later (99th percentile of a healthy reference population,
0.04 μg/L). After the second troponin measurement, an exercise stress echocardiogram is strongly positive at a low workload, with reproduction of the index symptoms at < 50% of the predicted workload and evidence of reversible ischaemia in the territory of the left anterior descending coronary artery. He has cardiac catheterisation the same day and is found to have a critical stenosis of the mid left anterior descending artery, which is treated with percutaneous coronary intervention and deployment of a drug-eluting stent. He makes an uneventful recovery, with normal left ventricular function.

3 Features associated with high risk, intermediate risk and low risk of adverse short-term outcomes in patients presenting with chest pain due to possible acute coronary syndrome

High risk

Presentation with clinical features consistent with acute coronary syndrome (ACS) and any of the following features:

Repetitive or prolonged (> 10 minutes) ongoing chest pain or discomfort
Elevated level of at least one cardiac biomarker (troponin recommended)
Persistent or dynamic electrocardiographic changes of ST-segment depression ≥ 0.5 mm or T-wave inversion ≥ 2 mm
Transient ST-segment elevation of ≥ 0.5 mm in more than two contiguous leads
Haemodynamic compromise
Sustained ventricular tachycardia
Syncope
Significant left ventricular (LV) dysfunction (LV ejection fraction < 40%)
Prior percutaneous coronary intervention within 6 months or prior coronary artery bypass surgery
Presence of diabetes or chronic kidney disease (estimated glomerular filtration rate < 60 mL/minute) and typical symptoms of ACS

Intermediate risk

Presentation with clinical features consistent with ACS and any of the following features, without high-risk features:

Chest pain or discomfort within the past 48 hours that occurred at rest, or was repetitive or prolonged (but currently resolved)
Age > 65 years
Known coronary artery disease
Prior aspirin use
Two or more of: hypertension, family history, current smoking, hyperlipidaemia
Presence of diabetes or chronic kidney disease and atypical symptoms of ACS

Low risk

Presentation with clinical features consistent with ACS without intermediate-risk or high-risk features. This includes onset of anginal symptoms within the past month, or worsening in severity or frequency of angina, or lowering of anginal threshold.

* Adapted from Box 8 in the National Heart Foundation of Australia and Cardiac Society of Australia and New Zealand Guidelines for the management of acute coronary syndromes 2006.6 Copyright 2006 The Medical Journal of Australia. Used with permission.

4 A proposed algorithm, incorporating an accelerated diagnostic protocol,* for assessment of possible cardiac chest pain after exclusion of ST-segment elevation on initial ECG

NHF/CSANZ = National Heart Foundation and Cardiac Society of Australia and New Zealand. ACS = acute coronary syndrome.
ECG = electrocardiogram. TIMI = Thrombolysis in Myocardial Infarction. * Based on the ADAPT study.11 † NHF/CSANZ Guidelines for the management of acute coronary syndromes 2006.6

5 Features of non-invasive tests available for further risk stratification of patients with chest pain, after excluding acute myocardial infarction

Procedural considerations

Relative contraindications*

Condition and test

Cost†

Radiation

Iodinated
contrast media

Inability to exercise

Significant resting ECG abnormality‡

Renal
impairment

Obesity

Severe airway disease

Myocardial ischaemia or perfusion

Exercise stress electrocardiography

Yes

Yes§

Stress echocardiography

No¶

Yes

No¶

Myocardial perfusion scanning

$$$

Yes

No¶

Yes

Yes§**

Obstructive coronary artery disease

Computed tomography coronary angiography

Yes

ECG = electrocardiogram. * Relative contraindications should be discussed further for individual patients. † Relative cost indications are based on current Medicare rebates. ‡ For example, left bundle branch block. § If there is significant functional impairment. ¶ If pharmacological stress testing can be performed. ** Adenosine is contraindicated.

6 Representative performance characteristics of non-invasive tests to identify myocardial ischaemia or obstructive coronary artery disease in patients with chest pain

Test	Sensitivity	Specificity

Exercise stress electrocardiography21	68%	77%
Stress echocardiography20	83%	77%
Exercise stress myocardial perfusion scanning19	85%–90%	70%–75%
Computed tomography coronary angiography22	99%	89%

The changing face of cardiovascular care in Australia

Realising the promised gains of cardiovascular innovation will demand an increase in clinical cardiac decision-making capacity

Modern cardiovascular care has been blessed with therapeutic innovations that have reduced morbidity and mortality. From invasive management for acute coronary syndrome (ACS) to nurse-led heart failure management programs and anticoagulation for stroke prophylaxis in atrial fibrillation (AF), therapeutic advances have led not only to improved cardiovascular outcomes, but also the need for greater sophistication in balancing benefits with harms and costs (Box). With such complexity, decisions about management increasingly require access to risk stratification and evidence-based clinical decision support.

In this issue of the Journal, we commence a series of articles aimed at reviewing current and future challenges in cardiac care, including new diagnostic and management approaches to chest pain, primary prevention strategies based on absolute cardiovascular disease risk, and acute and chronic care of coronary heart disease (CHD), AF and heart failure.

The evolving cardiovascular disease burden

Ongoing monitoring of cardiovascular outcomes by the Australian Institute of Health and Welfare (AIHW) has demonstrated reductions in cardiovascular mortality and morbidity. From 1968 to 2007, age-standardised death rates from cardiovascular disease fell by over 75%, from 1020 to 232 per 100 000 men and from 718 to 170 per 100 000 women.1 These gains exceed those observed in all other major diseases.2 Drawing on international data, the reduced rates have been attributed in about equal parts to effective primary prevention targeting hypertension, hypercholesterolaemia and smoking, and to more effective treatments for acute cardiac conditions, as well as improved secondary prevention and chronic disease management.3,4

However, a disproportionate burden of cardiovascular disease remains among people in rural and remote areas, those with linguistically and culturally diverse backgrounds, Indigenous peoples and older patients with multiple comorbidities. Geographic remoteness, communication difficulties, socioeconomic disadvantage, cultural disparities and limited evidence of intervention effectiveness are some of the factors that account for this uneven distribution.1,5

A rising prevalence of obesity and the associated increase in diabetes potentially herald a reversal in the declining rates of cardiovascular mortality. Between 1995 and 2007–08, the proportion of Australians aged 35–44 years classified as being overweight or obese increased by over 7 percentage points.1 Recent AIHW data suggest a plateau in previously observed declines in ACS events, although declines in CHD deaths are still evident. Registry data also show that about 50% of ACS events occur among patients with established CHD.6 These data reinforce the need for more widespread implementation of primary prevention targeting people at high absolute risk of CHD,7 combined with better secondary prevention in patients with established CHD.

Innovation in care

Diagnostic and therapeutic innovations in cardiovascular care come with increased costs attributable to the technology itself, plus those associated with training, recruitment and subsequent interventions. The value of these innovations lies in their promise to reduce deaths and morbidity, but their widespread uptake often occurs in the absence of robust evidence of benefit across the full spectrum of outcomes for patients and health services.8 Even when such evidence is available, their real-world value depends on effective implementation in cardiovascular health services. Generalisation of the benefits and risks observed in clinical trials of emerging pharmacotherapies and devices requires appropriate health service infrastructure to support effective adoption. New technologies will likely bring the need for accurate clinical risk assessment and customised management into sharper focus. For example, in investigating chest pain, routine use of high-sensitivity troponin assays or computed tomography coronary angiography in the emergency department is likely to lead to unacceptably high false-positive rates, and unnecessary investigations and treatment with little impact on patient outcomes.9 In light of the potential for harm from novel diagnostics and pharmacotherapies, evidence-informed patient selection is key to their clinical and cost-effectiveness. This is paramount, given the increasing numbers of patients with chronic disease, advancing age and multiple concomitant treatments.

The challenge of determining value for individual patients, by balancing potential benefit and potential harm, is common to most of the interventions in the Box. Transcatheter aortic valve implantation (TAVI)10 for severe aortic stenosis (AS) is a good example. As patients with this condition are often very old, and half die within 12 months of symptom onset, the challenge is to distinguish between patients who are dying from severe AS and those dying with it. Similarly, when choosing between TAVI and open aortic valve replacement (AVR), the risks of increased stroke and limited long-term clinical data for TAVI must be weighed against the risks of acute postsurgical complications from AVR in this older high-risk group.

Shifting paradigms of care

Effective transition of clinical decision making and care from the acute hospital setting to general practice and patient self-management are key to preventing avoidable cardiac events and readmissions. Transition-of-care initiatives include nurse-led heart failure management programs in the community that have reduced rates of rehospitalisation and mortality.11,12 Similar programs have potential in the management of AF and secondary prevention of CHD.13

Another paradigm shift is the development of integrated, statewide cardiac care networks, which seek to provide clinicians outside urban centres with rapid expert assessment of patients with acute cardiac conditions and prioritised transfer to tertiary centres. Federal health reforms associated with new funding arrangements and enhanced performance accountabilities will also act as a catalyst for further clinical service redesign to increase care access and integration. At a health service level, local hospital networks and Medicare Locals will facilitate collaborative implementation of multidisciplinary services that shift more of the care of patients with chronic cardiac conditions from the hospital to the community. This will demand a commensurate increase in clinical decision-making capacity in primary care.

All these changes will require better knowledge about the effectiveness, safety and accessibility of cardiovascular care in real-world settings14 by way of clinical and procedural registries, which allow benchmarking of clinical care and patient outcomes, rapid identification of evidence–practice gaps, and timely evaluation of benefits and harms of new technologies.15,16 The costs of maintaining nationally representative registries may be increasingly offset by automated data input from electronic health records and administrative datasets that quickly profile what, where and when cardiovascular care is provided. If implemented well, the timely collection of clinical data, and its standardisation using universal clinical definitions, will facilitate more accurate interjurisdictional comparisons.17 Public reporting of analyses of such data, conducted under the auspices of the AIHW and the Australian Commission on Safety and Quality in Health Care, should bring greater transparency, objectivity and accountability to cardiovascular care delivery. This enhanced data collection infrastructure should also facilitate applied clinical research that evaluates the comparative effectiveness of emerging and competing health care technologies and treatments, and constructs and validates new risk-prediction rules and management algorithms.

Continued investment by clinicians and health services in pursuing a “translatable” evidence base to inform decision making will be needed to ensure improved access to, and effective implementation of, cardiovascular innovations.

Uncertainties in implementation of emerging cardiac technologies

Innovation

Clinical dilemma

High-sensitivity troponin

What is its utility in ruling out acute coronary syndrome in patients presenting with chest pain?

What are the clinical significance and therapeutic implications of troponin elevation in heart failure, arrhythmias and non-cardiac diagnoses?

B-type natriuretic peptide

How can it be used in titration of therapies and disease monitoring in heart failure?

Computed tomography coronary angiography (CTCA) and
CTCA-based functional assessment

What is CTCA’s role in the early diagnostic work-up of undifferentiated chest pain?

Is CTCA cost-effective in the assessment of stable coronary artery disease?

Stress magnetic resonance imaging and three-dimensional echocardiography

Do these novel imaging modalities provide incremental diagnostic value beyond existing imaging modalities?

Electronic decision support

Does the integration of objective risk stratification with therapeutic recommendations lead to better care and clinical outcomes?

Examples of novel therapies:

Percutaneous implantation of aortic valves for aortic stenosis
Percutaneous technologies for mitral regurgitation
Implantable cardioverter defibrillators
Novel anticoagulants for stroke prophylaxis
Emerging pharmacotherapies for heart failure
Renal sympathetic denervation for management of hypertensive heart disease and refractory hypertension

Which patient characteristics identify those who will derive the greatest benefit from the specific therapy?

Which patient characteristics identify those most at risk of complications?

How can we objectively assess the balance between risk and benefit?

What are the downstream impacts of implementing the innovation?

What patient-specific outcomes are most relevant for assessing benefit and cost-effectiveness?

How do we develop a sustainable workforce for the provision of the specific therapeutic innovation?

Primary prevention of cardiovascular disease: new guidelines, technologies and therapies

Acontinuing trend in primary prevention of cardiovascular disease (CVD) in general practice has been the move away from managing isolated CVD risk factors, such as hypertension and dyslipidaemia, towards assessment and management of these factors under the banner of absolute CVD risk.1 This has been underscored by the publication of guidelines for assessment and management of absolute risk.2,3 These guidelines seek to consolidate various individual disease and risk factor guidelines, recognising CVD as a common end-disease pathway and, therefore, the benefit of taking a common absolute risk-based approach. The rationale behind adopting this approach can be summarised as follows:

Medication is best initiated in those most likely to benefit from it, and who therefore have a favourable risk-to-benefit ratio.
It is more cost-effective than intervention for single risfactors.
It avoids medicalisation of the low-risk population.
It better identifies those most likely to have covert CVD, avoiding costly additional investigations.
Beneficial therapeutic agents can be initiated at a level above the ideal rather than at an arbitrary cut point.
Due attention is paid to CVD risk, which might otherwise be subsumed within a particular chronic disease management strategy (eg, diabetes and blood glucose levels)4 (see example in Box 1).

Here, we provide information for general practitioners on new approaches to clinical management of CVD risk factors in patients without overt disease, and new technologies and therapies to assess and manage them.

The Australian absolute CVD risk guidelines

In the Australian National Vascular Disease Prevention Alliance (NVDPA) guidelines for assessing absolute CVD risk, absolute risk is calculated as the probability of a stroke, transient ischaemic attack, myocardial infarction, angina, peripheral arterial disease or heart failure occurring within the next 5 years.2 Absolute risk is categorised, and can be communicated to patients, as low (< 10%), moderate (10%–15%) or high (> 15%). Medication is recommended for individuals at high risk and sometimes for those at moderate risk if additional risk factors are at play (eg, Aboriginal or Torres Strait Islander status or a family history of premature CVD).

Doctors can reliably estimate relative risk — that is, the risk level of an individual with a risk factor compared with someone who does not have that risk factor.5 The problem with relative risk is that it tells you that a smoker is at greater risk than a non-smoker but does not convey what that risk actually is. The absolute CVD risk calculator recommended by the NVDPA (http://www.cvdcheck.org.au) is based on the Framingham Heart Study.2,6,7 It has good predictive value for subsequent CVD events in untreated individuals and has been validated in the Australian population aged 30–74 years.8

If a patient has manifest CVD (eg, past history of stroke or myocardial infarction) or already has a condition that places him or her at high risk of CVD (Box 2), then no risk assessment is required before commencing blood pressure (BP)-lowering and/or lipid-lowering therapy. The NVDPA assessment guidelines recommend that all other adults aged 45–74 years should be assessed for cardiovascular risk.2 Below the age of 45 years, almost all patients will be at low risk. For people older than 74 years, the guidelines recommend entering their age as 74 in the calculator, to provide a minimum estimate of risk.

All attempts at recalibrating calculators for Aboriginal or Torres Strait Islander peoples have so far failed, with recognition that risk is underestimated in this population.9 Assessment should commence in Aboriginal or Torres Strait Islander adults at the age of 35 years (in deference to the reduced life expectancy in this population) and the score used as a minimum estimate of risk.

Barriers to uptake of the absolute risk approach, such as acceptability, feasibility and effectiveness in the primary care context, need to be overcome.10 Depending on the clinical context, treatment on the basis of elevated single risk factors may still be appropriate. For example, atrial fibrillation has a well recognised thromboembolic stroke risk, which warrants a disease-specific stroke and bleeding risk assessment for anticoagulant or antiplatelet therapy.

Recommended changes to clinical practice from the 2012 management guidelines

The 2012 NVDPA guidelines for managing absolute CVD risk recommend both BP-lowering and lipid-lowering agents for all patients at high absolute risk of CVD, unless contraindicated or clinically inappropriate.3 For patients at moderate risk, treatment with a BP-lowering and/or lipid-lowering agent should be considered if the risk remains elevated after lifestyle interventions, BP is ≥ 160/100 mmHg, there is a family history of premature CVD, or the patient is of South Asian, Middle Eastern, Maori, Pacific Islander, Aboriginal or Torres Strait Islander ethnicity. The guideline authors recommend that people with a BP of ≥ 160/100 mmHg be treated for their BP regardless of their absolute risk level. The 2012 guidelines have also revised and simplified BP targets to aim for with BP-lowering treatment and lifestyle measures: for the general population or those with a reduced glomerular filtration rate, the target is ≤ 140/90 mmHg; and for people with microalbuminuria, macroalbuminuria or diabetes, the target is ≤ 130/80 mmHg.

Aspirin and other antiplatelet agents are no longer routinely recommended for use in primary prevention of CVD, including for people with diabetes or high absolute CVD risk. Previous recommendations for people with diabetes were based on the assumption of equivalent CVD risk to those with established CVD but without diabetes.11 However, recent primary prevention trials in patients with diabetes have not shown benefit for aspirin.12,13 Harm–benefit analyses of antiplatelet drugs for primary prevention assume that risk of CVD rises with age but risk of adverse effects does not. While it is true that CVD risk is largely determined by age, the risk of adverse effects is also likely to be higher in older people.14 An ongoing clinical trial, Aspirin in Reducing Events in the Elderly (ASPREE), is being conducted in Australian general practice to examine whether the benefits of routine aspirin outweigh the harms in patients aged 70 years or older.15

Lifestyle interventions

Regardless of a patient’s risk level, the advice in the 2012 NVDPA guidelines remains that treatment should always begin with lifestyle interventions, such as smoking cessation; reducing intake of dietary salt, fat, high-calorie drinks and overall calories; and increasing exercise. Most people at moderate absolute risk should be given the opportunity to reduce their risk by following lifestyle advice, with drug therapy only considered if their risk does not reduce in 3–6 months or if they have specific additional risk factors, such as Aboriginal or Torres Strait Islander status or a family history of premature CVD.

Smoking is the most important modifiable risk factor, and action on smoking is always the highest-priority lifestyle intervention. Smoking cessation reduces the risk of CVD substantially and sustainably, and it also reduces all-cause mortality.16 Health professional advice, nicotine replacement therapy and medication are effective smoking cessation interventions.17–19

Weight loss is important in that it reduces the risk of elevated BP and lipid levels and diabetes. Even modest weight loss (5%–10% of initial weight) can improve health.20 There are no simple answers to the question of which diet will achieve weight loss. Whichever diet is chosen, it needs to be sustainable to be effective. There is some evidence that low-carbohydrate–high-protein diets, such as the CSIRO (Commonwealth Scientific and Industrial Research Organisation) diet, have greater weight loss and lower attrition rates in the short term, but longer-term evidence is lacking.21

Weight-loss medications available to date have been disappointing because of the lack of sustained weight loss and the risk of side effects. Several weight-loss medications have been withdrawn from market due to harmful effects, the most recent being sibutramine.22 In addition, weight loss achieved using medication is unlikely to have the same health benefits as weight loss achieved by diet and exercise, with all their associated benefits for health and wellbeing. The draft National Health and Medical Research Council Clinical practice guidelines for the management of overweight and obesity in adults, adolescents and children in Australia23 recommend orlistat as an agent with proven effectiveness in adults,24 although its use will be limited by the acceptability of side effects, such as flatulence and anal leakage.

Weight-loss surgery has shown promise for patients with significant obesity. The Swedish Obese Subjects study found average weight loss from various types of bariatric surgery of 14%–25% over 10 years, and a reduction in all-cause mortality, diabetes and CVD.25 However, this was not a randomised controlled trial, and the intensity of monitoring and follow-up of patients may influence the generalisability of the study results. Weight-loss surgery is recommended if a patient has a body mass index > 40 kg/m2, or > 35 kg/m2 with comorbidity.26

Regular physical activity reduces CVD risk and individual CVD risk factors and protects against other diseases.3 Health benefits are achieved with around 150–300 minutes of moderate-intensity activity or 75–150 minutes of vigorous activity each week.3,23

How do guideline recommendations align with prescribing criteria for lipid-lowering drugs?

In 2006, the Pharmaceutical Benefits Advisory Committee (PBAC) revised the Pharmaceutical Benefits Scheme (PBS) General statement for lipid-lowering drugs prescribed as pharmaceutical benefits.27 This revision aimed to bring the PBS prescribing criteria for lipid-lowering drugs more in line with the absolute risk approach, while recognising that, at the time, a lack of widespread access to a CVD risk calculator was a barrier to using absolute risk as a prescribing criterion. Conditions considered in the NVDPA guidelines2,3 to confer a high risk of CVD that are not currently included in the PBS criteria are: moderate or severe chronic kidney disease; total cholesterol level > 7.5 mmol/L in males who are less than 35 years old and in premenopausal women; and systolic BP ≥ 180 mmHg and total cholesterol level < 6.5 mmol/L, or total cholesterol level < 5.5 mmol/L and high-density lipoprotein cholesterol level > 1.0 mmol/L.

To date, what has not been presented to the PBAC for consideration is the effectiveness and cost-effectiveness of lipid-lowering treatments for patients at high (or moderate) absolute risk with “normal” lipid levels. From our previous analysis of the Australian Diabetes, Obesity and Lifestyle Study (AusDiab) cohort, about 90% of patients at high absolute risk who do not meet the current PBS criteria for prescription of lipid-lowering drugs are in this category.28

How frequently should absolute CVD risk
be monitored?

Once management decisions have been made, absolute risk should be monitored according to the recommendations in Box 3. The reassessment of absolute risk in the absence of a trigger such as initiation of smoking or diabetes diagnosis may be conducted at longer intervals than currently recommended, especially in low-risk individuals, as reclassification (ie, moving from low to moderate or moderate to high risk), which would lead to management changes, is likely to be an infrequent phenomenon.29 If a patient is already being treated for elevated BP or lipid levels, the pretreatment values should be used to calculate absolute risk.

Is there evidence for the absolute CVD
risk approach?

Using the absolute risk approach, patients who have isolated elevated risk factors, but low absolute risk, will generally not be treated with medication. Because age is such a strong predictor of risk, this means that younger patients with isolated elevated risk factors will in general not be treated with BP-lowering or lipid-lowering agents. Many clinicians may be uncomfortable with this approach, as they feel that delaying treatment until it reaches a particular risk threshold can allow irreparable damage to occur. It is unlikely that there will be a randomised controlled trial of the absolute risk approach versus the isolated risk factor approach to test this, because of the sample size and time that would be required.

However, previously conducted trials support the absolute risk approach. Individual patient data (IPD) meta-analyses of both BP-lowering and lipid-lowering drug trials have shown that the relative risk reduction of cardiovascular events is consistent regardless of baseline BP or lipid levels. The IPD meta-analysis of BP-lowering drug trials showed that the relative risk reduction was constant down to the lowest BP levels observed in the trials (110 mmHg systolic and 70 mmHg diastolic), and that results were consistent in trials of patients with a prior history of coronary heart disease or stroke and those with no prior history of vascular disease.30 The same result has been observed in cohort studies.31 Similarly, the recently updated IPD meta-analysis of lipid-lowering drug trials by the Cholesterol Treatment Trialists’ Collaboration confirms that the relative risk reduction is consistent in patients with or without pre-existing CVD and is independent of the baseline cholesterol level.32 This study provided further empirical evidence to support the absolute risk approach, by showing a constant relative risk reduction regardless of the baseline risk of a cardiovascular event, and therefore increasing benefits from treatment in patients with increased absolute risk of CVD.

New technologies for CVD risk factors and
risk assessment

New technologies that are currently having an impact on BP management in general practice are ambulatory BP devices and oscillometric BP devices (for both clinic and home use).33,34 These devices permit an estimation of BP that is more representative of usual BP and associated CVD risk, through a combination of reducing “white coat” effects and observer error, allowing systematic collection of multiple BP recordings and, for measures made outside the clinic, identification of masked hypertension. Ambulatory BP monitoring involves measuring BP at regular intervals over a 24-hour period while patients undergo normal daily activities, including sleep. Home BP monitoring is a validated method for monitoring and managing a patient’s BP, which can be readily incorporated into practice. Where barriers to ambulatory and home BP monitoring exist, oscillometric devices can be used to approximate mean daytime ambulatory BP.35 This “automated office BP” measurement has three basic principles: multiple BP readings are taken; an automated device is used; and measurements are taken while the patient rests quietly alone. The oscillometric device distributed by the High Blood Pressure Research Council of Australia can be used in this way. The machine can be set to automatically record three BP measures at 5-minute intervals. The patient is then left to sit alone for 15 minutes in a room or a screened area, and the BP value displayed after this time is the average of all recordings. These devices can also be used as a screening device for peripheral arterial disease.36 Users should be aware that BP levels measured this way are generally 5 mmHg lower than clinic measures.37

Multiple clinical, biomarker and imaging tests have been proposed as methods for identifying patients at high risk of CVD. Of most clinical use would be tests that could more effectively discriminate moderate-risk patients who are actually at high or low risk of a cardiovascular event. A series of recent reviews has shown that many of the studies aiming to improve identification of patients at increased risk by using non-traditional risk factors, such as C-reactive protein (CRP) and fibrinogen, had methodological flaws and that, on the evidence to date, these factors were unlikely to improve the discrimination of risk.38 Similarly, using apolipoproteins A and B reclassifies less than 1% of patients beyond the classification based on traditional risk factors.39 The cost, inconvenience to patients and potential harm mean that calls for these tests to be used more widely are premature. Biomarkers already in use in general practice, such as CRP, add very little to current risk algorithms.40 The use of computed tomography coronary angiography to screen patients needs careful evaluation of cost and radiation risk before implementation.41 Coronary artery calcium scoring may have a future role in reclassification for individuals found to be at moderate risk using routine risk stratification.42

New therapies for CVD risk factors

People who regularly consume fish have lower CVD event rates than non-consumers.43,44 However, intervention trials of fish oils are less convincing. A meta-analysis of 48 randomised controlled trials showed no benefit of omega-3 fats on mortality or cardiovascular events in patients with or without existing coronary heart disease.45 Therefore, in primary prevention, it is justified to recommend the consumption of fish as part of a healthy diet, without the need to use fish oil supplements.

Denervation of the kidney using minimally invasive devices has BP-lowering effects in the majority of treated individuals, but it may also have benefits for glucose metabolism, renal impairment, left ventricular hypertrophy, and other conditions.46 This method is still early in its development and availability.

Conclusion

The move to an approach based on absolute risk for the primary prevention of CVD is likely to improve the effectiveness and cost-effectiveness of treatment, and the 2009 and 2012 NVDPA guidelines support this approach. The absolute risk approach targets the patients who are most likely to benefit from medication, and reduces the medicalisation of patients at low risk. The increasing availability of cardiovascular risk calculators, either on the internet or as standalone software, also removes one of the barriers to implementing the absolute risk approach. New technologies have varying evidence of utility, but oscillometric BP devices can be readily adopted. The role of coronary artery calcium scoring and other biomarkers in risk stratification is yet to be established.

1 Case example: how the absolute risk approach better targets therapy

Joe is a 64-year-old man who smokes but does not have diabetes
or known cardiovascular disease (CVD). His blood pressure is
136/82 mmHg, total cholesterol level is 5.4 mmol/L and high-density lipoprotein (HDL) cholesterol level is 1.0 mmol/L.

Jane is a 46-year-old woman who does not smoke and does not have diabetes or known CVD. Her blood pressure is 142/82 mmHg, total cholesterol level is 6.5 mmol/L and HDL cholesterol level is 1.4 mmol/L.

Using the isolated risk factor approach, other than smoking, Joe has
no elevated individual risk factors that would warrant treatment with medication. Jane, on the other hand, has hypercholesterolaemia and hypertension that would see her taking lifelong antihypertensive and lipid-lowering therapy.

However, using the absolute risk approach, Joe’s absolute risk is high (21%) and Jane’s is low (3%). Joe requires medication in addition to lifestyle changes, while Jane needs attention paid to her antecedent risk behaviour rather than medication.

2 Conditions conferring a high risk of cardiovascular disease*

Diabetes and age > 60 years
Diabetes with microalbuminuria (> 20 µg/min or urinary albumin : creatinine ratio > 2.5 mg/mmol for males, > 3.5 mg/mmol for females)
Moderate or severe chronic kidney disease (persistent proteinuria or estimated glomerular filtration rate
< 45 mL/min/1.73 m2)
A previous diagnosis of familial hypercholesterolaemia
Systolic blood pressure ≥ 180 mmHg or diastolic blood pressure ≥ 110 mmHg
Serum total cholesterol level > 7.5 mmol/L

3 Recommended frequency of monitoring for absolute cardiovascular disease (CVD) risk*

Regular review of absolute CVD risk is recommended at intervals according to the initial assessed risk level:

Regular review of absolute CVD risk is recommended at intervals according to the initial assessed risk level:
Low (< 10% risk of cardiovascular event within 5 years): review every 2 years
Moderate (10%–15% risk of cardiovascular event within
5 years): review every 6–12 months
High (> 15% risk of cardiovascular event within 5 years): review according to clinical context

J-curve revisited: cardiovascular benefits of moderate alcohol use cannot be dismissed

There is universal agreement that the alcohol-related problems of violence, motor vehicle fatalities and injuries, family disruption and significant ill health are major challenges for the Australian community,1,2 as they are worldwide.3 Concerted action is required to address this.4 In response, Australia’s National Health and Medical Research Council (NHMRC) produced a thoroughly researched document in 2009,5 which updated its 2001 Australian alcohol guidelines.6

The NHMRC guidelines, important as a basis for alcohol policy in Australia, were unbalanced on an important issue: the potential health benefits of one to two standard drinks per day were reported incompletely. In a 118-page report, the potential health benefits of low doses of alcohol were dealt with in less than half a page. It made no comment on the extent of the literature showing that people consuming one to two standard drinks per day have a lower risk of cardiovascular outcomes than abstainers. At the time of the 2009 NHMRC guidelines, there had been over 60 high-quality large-scale studies on alcohol use on over a million subjects and associations with over 94 500 deaths.7,8 This extensive database has widely influenced alcohol policy in many countries and has been the basis of statements on alcohol intake and cardiovascular disease from most national and international health advisory bodies (Box 1).

The NHMRC guidelines appeared to dismiss this extensive evidence base with the statement that “some recent studies have found no significant cardioprotective or all-cause associations” and suggested that any effect may have been overestimated. It relied for this assessment on a single publication that queried the prior research on the apparent cardioprotective effect of low to moderate alcohol intake.9 It has already influenced other Australian bodies in their alcohol policy; it is repeated without any update from recent literature in the 2011 position statement of the Cancer Council of Australia, which highlights the potential for low doses of alcohol to cause cancer.10

The dismissal of the cardiovascular and overall mortality benefits of low to moderate intake of alcohol has the potential to bias the debate about the dangers of alcohol in our community. The recent campaigns to publicise the potential for alcohol to cause cancer11,12 have been launched without providing balancing information that drinking alcohol within the limits set by the NHMRC guidelines may have cardiovascular and overall mortality benefits.

The J-shaped curve

Multiple studies from the late 1970s13 onwards have reached a general consensus: people consuming one to two standard drinks per day appear to have a lower cardiovascular event rate than persons abstaining from alcohol; a relationship described as a J-shaped (or U-shaped) curve.14 Estimates of the risk reduction associated with one to two standard drinks per day compared with those abstaining from alcohol have ranged from 20% to 30%.7 Over many years, there have been criticisms of the interpretation of this association, the two most prominent being the possibilities of misclassification of the reference group and confounding from uncontrolled other lifestyle factors. This brief review makes no comment on the important issue of whether the association can be construed as a causal relationship. This question is unlikely to be resolved without a randomised trial.15 Many feasible biological pathways for a possible mechanism of benefit of alcohol have been proposed,16 and beneficial trends in biomarkers with low to moderate intake of alcohol strengthen the case17 but do not prove causation per se. The main focus of this review is whether misclassification of drinking habits explains the J-shaped relationship between alcohol consumption and cardiovascular disease.

Misclassification of drinking habits

In stating that “some recent studies have found no significant cardioprotective or all-cause associations”, the 2009 NHMRC report relied on a 2006 meta-analysis by Fillmore and colleagues9 and an expansion of their views in a 2007 review.18 Their meta-analysis presented a reinterpretation of previous studies, developing and expanding the “sick quitter” hypothesis first developed in 1988 by Shaper and colleagues.19 This interpretation broadly proposes that misclassification of drinking habits may have resulted in more coronary events in the abstainer reference group because it included ex-drinkers with an increased cardiovascular risk because of ill health or increasing age. The sick quitter hypothesis was criticised in several studies carried out during the 1990s. Fillmore and colleagues re-examined the hypothesis using their own methodology to exclude all studies with misclassification errors in the abstainer reference group. However, they took such a strict approach that they identified only seven studies on total mortality and two on coronary heart disease mortality free of the misclassification errors. In this small number of studies, the evidence base remaining failed to confirm the J-shaped relationship between alcohol consumption and cardiovascular events. The conclusion from their analysis was that “future prospective studies should use far greater precision in their assessment of drinking behavior and abstinence”.18 Recent studies confirm the suspicion that that there is indeed a difference between former drinkers and lifetime abstainers in their pattern of ischaemic heart disease (IHD) incidence and mortality, but no evidence that this differentiation makes a substantive difference to the relationship between lower levels of cardiovascular events with a low to moderate intake of alcohol.20 While it is possible that younger adults with a disability may choose an abstinent lifestyle and suffer a higher mortality in adulthood from their disability and not their abstinence, this accounts for only a very small proportion of non-drinkers and is unlikely to contribute substantially to the J-shaped relationship.21

Since the misclassification version of the sick quitter hypothesis was put forward in 2006 and embraced in the 2009 NHMRC guidelines, the relationship and the possibility of misclassification in the abstainer group has been re-examined in several recent major meta-analyses. Corrections for possible misclassification error have been attempted in more recent studies, and none support misclassification as an interpretation of the J-shaped curve.

Patients with incident cardiovascular disease

Three recent meta-analyses on incident cardiovascular disease do not support the misclassification hypothesis. A BMJ meta-analysis published in late 2011 represents the most complete meta-analysis to date.22 It found no differences in the extent of relative risk reduction in cardiovascular disease mortality when classification adjustments were made to address the sick quitter misclassification hypothesis. Of the 4235 studies considered and 84 studies involving over one million people included in the final analysis, the pooled estimates showed a lower risk of all-cause mortality for drinkers compared with non-drinkers (relative risk, 0.87; 95% CI, 0.83–0.92). The meta-analysis did not include a graphic representation of the J-curve, but it did report the dose relationships that can be graphed as shown in Box 2, confirming the J-shaped relationship between alcohol intake and cardiovascular morbidity and mortality. A variety of cardiovascular outcomes were examined. The relative risks were no different when active low to moderate drinkers were compared with all abstainers versus lifetime abstainers. The relative risk of experiencing a cardiovascular outcome among those with a light to moderate intake when the reference group was limited to lifetime abstainers was 0.82 (95% CI, 0.78–0.86). The relative risks for incident coronary heart disease were even lower.

This meta-analysis has been criticised by Fillmore and colleagues for including many of the studies that they had previously ranked as statistically inadequate,23 but two other more recent meta-analyses have specifically addressed the question of misclassification. An analysis comparing lifetime abstainers with former drinkers showed that inclusion of former drinkers in the reference group may influence estimates of IHD morbidity, but not estimates of IHD mortality.20 Another meta-analysis, which completely excluded any studies that may have included former drinkers or sick quitters and restricted itself to lifetime abstainers, assessed 38 627 IHD events (mortality or morbidity) among 957 684 participants.24 This thorough analysis showed differential risk curves for sex and end point, but the overall conclusion was that there was a cardioprotective effect with low to moderate alcohol intake. The trends for men and women at the low to moderate end of the scale of daily consumption presented in this analysis are shown in Box 3. When all levels of consumption were analysed with lifetime abstainers as the reference group, the typical J-shaped pattern emerged (Box 2).

Patients with established cardiovascular disease

Studies of people who already have cardiovascular disease do not support the misclassification hypothesis. In 2010, a large meta-analysis examined the effect of alcohol consumption in patients with established cardiovascular disease, finding no difference in its conclusions when studies with lifetime abstainers were analysed separately from all abstainers.25 Seven studies comprising 12 819 patients with cardiovascular disease provided detail on dose–response relationships between alcohol consumption and cardiovascular outcomes. This analysis confirmed the J-shaped relationship for alcohol intake for patients with established cardiovascular disease.

Regular low to moderate alcohol intake compared with binge drinking

The J-shaped association between alcohol consumption and cardiovascular disease applies to regular low to moderate alcohol intake but not to binge drinking. Two recent meta-analyses have demonstrated a lower risk of cardiovascular disease with a regular daily intake of a low to moderate dose of alcohol and a higher risk with infrequent binge drinking.26,27

Confounding and causation

This brief review is an update of the evidence on whether a regular low to moderate intake of alcohol is associated statistically with a lower risk of coronary heart disease. It does not comment on whether the lower risk implies that a causal relationship is simply because of lifestyle factors unrelated to alcohol intake. While coronary heart disease is the only disease that shows the J-shaped relationship with alcohol,28 the question of whether a low to moderate intake of alcohol is truly protective is unlikely to be resolved without a randomised controlled trial.15 While many explanations for a possible mechanism of benefit of alcohol in coronary heart disease have been proposed,16 they do not prove a causal relationship per se. Extensive meta-analysis of 21 biomarkers strengthened the case for a causal relationship, concluding that favourable changes in cardiovascular biomarkers “provide indirect pathophysiological support for a protective effect of moderate alcohol use on coronary heart disease”.17 While the observational evidence is inconclusive that a low to moderate intake of alcohol is cardioprotective, much of the data on the adverse health effects of alcohol depend on a similar quality of evidence, including the conclusion that it is a carcinogen.29

Conclusion

The evidence for a J-shaped relationship between alcohol consumption and cardiovascular outcomes is extensive. The possibility that the shape of the curve may be due simply to a misclassification of drinking habits, as suggested in the 2009 NHMRC guidelines, has not been supported by recent studies. The overall impact on community health of a 20%–30% lower rate of coronary disease in light to moderate drinkers is likely to be substantial.30 While the J-shaped curve complicates the formulation of public policy on alcohol,31 it cannot be dismissed.

1 Statements from international medical bodies on the relationship between low to moderate alcohol intake and cardiovascular disease

Organisation

Relevant quote

Year

Reference

American Heart Association

“There are more than 60 prospective studies that suggest an inverse relation between moderate alcoholic beverage consumption and coronary heart disease”

2001

http://my.americanheart.org/professional/StatementsGuidelines/ByTopic/TopicsA-C/Statement-Guideline-Topics-A C_UCM_322827_Article.jsp (accessed Mar 2013)

Royal Australasian College of Physicians

“The injury and chronic disease burden associated with risky or high risk levels of alcohol consumption are offset by a decrease in the burden of cardiovascular disease produced by minimal alcohol consumption”

2005

Royal Australasian College of Physicians. Alcohol policy: using evidence for better outcomes. Sydney: RACP, 2005

American Diabetes Association

“In individuals with diabetes, light to moderate alcohol intake (1 or 2 drinks per day; 15 to 30 g alcohol) is associated with a decreased risk of CV disease, which does not appear to be due to an increase in HDL cholesterol”

2006

Bantle JP, Wylie-Rosett J, Apovian CM, et al. Nutrition recommendations and interventions for diabetes — 2006: a position statement of the American Diabetes Association. Diabetes Care 2006; 29: 2140-2157

British Heart Foundation

“Moderate drinking, that is 1 or 2 units a day, may offer some protection from coronary heart disease”

Current

www.bhf.org.uk/alcohol (accessed Mar 2013)

World Health Organization

“Light to moderate drinking can have a beneficial impact on morbidity and mortality for ischaemic heart disease and ischaemic stroke”

Current

http://www.who.int/substance_abuse/publications/global_alcohol_report/msbgsruprofiles.pdf (accessed Mar 2013)

European Society of Cardiology

“recent meta-analyses carried out in large series of subjects and published in prestigious journals show that 1–2 drinks per day may be in fact, beneficial”

Current

http://www.escardio.org/communities/councils/ccp/e-journal/volume9/Pages/low-doses-of-alcohol-Clement-Denis.aspx (accessed Mar 2013)

2 Meta-analysis showing the J-shaped relationship between cardiovascular mortality and alcohol intake based on 84 studies involving over a million people

Redrawn from data in Ronksley et al.22

3 Lifetime abstainers compared with occasional and low to moderate consumers of alcohol, showing associations with lower rates of ischaemic heart disease (IHD) morbidity and mortality in men and reduced morbidity in women

Redrawn from data in Roerecke and Rehm.24

Screening, referral and treatment for depression in patients with coronary heart disease

In 2003, an Expert Working Group of the National Heart Foundation of Australia (NHFA) issued a position statement on the relationship between “stress” and heart disease. They concluded that depression was an important independent risk factor for first and recurrent coronary heart disease (CHD) events.1 Here, we provide an update on evidence obtained since 2003 regarding depression in patients with CHD, and include guidance for health professionals on screening and treatment for depression in these patients. Our statement refers to depression in general (mild, moderate and severe), as all grades of depression have an impact on CHD prognosis. The process for developing this consensus statement is described in Box 1. Treatment decisions should take into account the individual clinical circumstances of each patient.

Epidemiology

The prevalence of depression is high in patients with CHD. Rates of major depressive disorder of around 15% have been reported in patients after myocardial infarction (MI) or coronary artery bypass grafts.3,4 If milder forms of depression are included, a prevalence of greater than 40% has been documented.3,4 Recently, the EUROASPIRE III study investigated 8580 patients after hospitalisation for CHD.5 The proportion of patients with depression, measured by the Hospital Anxiety and Depression Scale, varied from 8.2% to 35.7% in men and 10.3% to 62.5% in women. This is consistent with Australian and New Zealand data from a 6-year study, Long-term Intervention with Pravastatin in Ischaemic Disease (LIPID).6,7 At the end of this trial, 27% of men and 35% of women were identified as depressed, using the Beck Depression Inventory II (BDI-II) questionnaire.

A large systematic review in 2006 suggested that individuals with depression, but no current CHD, have a moderately elevated risk of 1.6 for a later index CHD event.8 This elevated risk was confirmed in the Whitehall II study of 5936 healthy individuals over a 6-year period, in which depression was associated with a hazard ratio of 1.93 for cardiovascular events.9 In the Nurses Health Study, 78 282 healthy women were assessed for depression. In the 6-year follow-up period, 4654 deaths were reported, including 979 deaths from cardiovascular disease.10 Depression was associated with increased all-cause mortality, with an age-adjusted relative risk of 1.76 (95% CI, 1.64–1.89).10 The effect of depression on CHD incidence is thought to be strongest around the time of the depressive episode, with longer-term effects mediated via recurrence of depression.11 In young people the association between depression and CHD may be stronger.12

The case–control INTERHEART Study included 11 119 patients with MI from 52 countries.13 Perceived stress and depression were shown to be important risk factors, which together accounted for 32.5% of the population attributable risk (PAR) for CHD, suggesting that together they were as important as smoking and more important than diabetes (PAR, 9.9%) and hypertension (PAR, 17.9%) as risk factors for CHD.13

For people with CHD and comorbid depression, the relative risk (RR) of death is increased (RR, 1.80 [95% CI, 1.50–2.15]), independent of standard risk factors for secondary prevention.8 Comorbid depression also leads to a higher risk of other adverse outcomes in patients with CHD, such as a lower likelihood of return to work, poorer exercise tolerance, less adherence to therapy, greater disability, poorer quality of life, cognitive decline and earlier dependency.14–20 Individuals with CHD and comorbid depression often have less access to interventions for CHD, despite being in a higher-risk group.21–23

Definition of depression and types of depression

The diagnosis of depression can be difficult in people with cardiovascular disease, as depressive symptoms such as fatigue and low energy are common to both CHD and heart failure, and may also be a side effect of some drugs used to treat cardiovascular disease, such as β-blockers.24 The diagnosis may be further complicated in such patients by their responses to their disease (and the associated stigma), which may include denial, avoidance, withdrawal and anxiety.

According to the Diagnostic and statistical manual of mental disorders, fourth edition (DSM-IV),25 major depression is diagnosed when there is a minimum of 2 weeks of depressed mood and/or lack of pleasure (anhedonia), accompanied by four or more other (listed) symptoms such as sleep disturbance, appetite disturbance, poor energy, psychomotor impairment or agitation, poor concentration or poor decision making, and suicidal ideas or thoughts of death. The association with CHD appears to increase with greater severity of depressive symptoms across the spectrum, with no discrete cut-off point at “major depression”. Some studies have suggested links between particular subtypes of depression, such as somatic or anhedonic depression, but these are not consistent findings.26–28

Screening for depression in patients with coronary heart disease

Screening of a population group for a risk factor or disease is worthwhile when the risk factor or disease has a reasonably high prevalence, there is a robust screening test, and effective and cost-effective treatments are readily available.29,30 Depression is both a risk factor and a disease in its own right, and fulfils these criteria for population screening. Screening for depression in patients with CHD would be expected to produce a higher yield than screening for depression in the general population, owing to a much higher prevalence of depression in patients with CHD. It is important to recognise depression in patients with CHD in order to provide the best possible care. Asymptomatic patients with significant cardiovascular risk factors (eg, those with diabetes) may also be considered for screening, as they have a high risk of depression.31

Many self-reported screening tools exist with the aim of detecting possible depression. These include the Patient Health Questionnaire (PHQ-2, PHQ-9), the Cardiac Depression Scale (CDS), the BDI-I and -II, and the Hospital Anxiety and Depression Scale.32,33 The BDI appears to be the most commonly used tool in studies involving cardiac patients. The CDS was developed by a member of the Expert Working Group (D L H) specifically for patients with cardiac disease.33 The short version (short form) has only five items. There is limited but expanding information on the use of the PHQ-9 in patients with cardiac disease.34,35 It is used widely in primary care. Simple tools such as the Kessler Psychological Distress Scale (K10),36 a measure of general distress, will often overdiagnose depression. This tool is currently used in mental health plans in Australia; however, there is no evidence of its use specifically for patients with CHD.

Recognising the need for a simple screening tool for depression in cardiovascular patients, the 2008 American Heart Association (AHA) Science Advisory suggested the use of the PHQ-2.37 The PHQ-2 is an abbreviated form of the PHQ-9, with only the first two of the nine questions in the PHQ-9 (Box 2).38 There are also other versions of the PHQ-2, which may use shorter time frames. The AHA recommended the use of the PHQ-9 if depression was noted using the PHQ-2.37 The Royal Australian College of General Practitioners’ Guidelines for preventive activities in general practice (the red book) also uses a categorical (Yes/No) version of the PHQ-2.39

The PHQ-2 and the PHQ-9 screening tools are associated with reasonable sensitivity and specificity.34 Importantly, depression diagnosed with the PHQ-2 and the PHQ-9 has been shown to predict worse CHD outcomes. In the Heart and Soul Study, positive responses to either question in the PHQ-2 (Yes/No version) predicted a 55% greater risk of cardiovascular events.35 Furthermore, the validity of the PHQ-2 and the PHQ-9 has been assessed in a variety of patients with varying clinical problems, ages and ethnicities, including in Australian Aboriginal people from urban and rural areas and people from the Torres Strait Islands.40–42 Adapted versions exist for use with Indigenous people.

Access to each of the tools varies. No copyright is breached by use of the PHQ-2 and the PHQ-9, and the PHQs and the CDS are free.43 However, some questionnaires such as the BDI-I and -II are subject to copyright and a royalty must be paid each time they are used.44

Implicit in the AHA Science Advisory37 is that screening and identification of patients with depression leads to appropriate treatment, or referral for treatment, by the responsible attending medical practitioner. Unfortunately, research has shown that screening may have little or no impact on the treatment of depression or on outcomes.45,46 Screening by nurses, researchers, receptionists or social workers is not sufficient without appropriate referral or treatment.

It is recommended that a simple tool, such as the PHQ-2 or the short-form CDS, is incorporated into routine screening of patients with CHD. Routine screening for depression is indicated at first presentation, and again at the next follow-up appointment. A follow-up screen should occur 2–3 months after a CHD event. Screening should then be considered on a yearly basis, as for any other major risk factor for CHD. Consideration should also be given to screening the partner or spouse of these patients for depression, as studies show that they are at an increased risk of developing depression.47 If screening is followed by comprehensive care, depression outcomes are likely to be improved.

Treatment of depression in patients with CHD

Collaborative care

Although individual treatment approaches and strategies have been studied, in practice a collaborative-care or stepped-care approach is probably optimal for managing patients with CHD and comorbid depression. The concept of collaborative care involves a group of health professionals working together in a coordinated manner, and this approach has consistently been shown to be associated with greater improvement in depression for patients with CHD compared with standard care, and to be cost-effective.48–52 For example, collaborative care after coronary artery bypass grafting improved depression scores, but not physical function or re-hospitalisation rates.48 In patients with depression comorbid with poorly controlled diabetes and/or CHD, the collaborative approach resulted in improvement in depression scores, glycated haemoglobin levels, low-density lipoprotein cholesterol levels, and systolic blood pressure.50

The Coronary Psychosocial Evaluation Studies (COPES) trial52,53 used a stepped-care treatment approach in patients with acute coronary syndromes (ACS) and persistent depression. Depressive symptoms decreased substantially in the intervention (stepped-care) group. Only three (4%) of the intervention patients experienced major adverse cardiac events compared with 10 (13%) of the patients given usual care, suggesting improved cardiovascular outcomes. Moreover, this stepped-care approach was associated with a 43% lower total health cost over the 6-month trial period.54

Pharmacological therapy

The efficacies of fluoxetine,55 sertraline (Sertraline Antidepressant Heart Attack Randomized Trial [SADHART],56 Enhancing Recovery in Coronary Heart Disease Patients [ENRICHD] trial),57 citalopram (Cardiac Randomized Evaluation of Antidepressant and Psychotherapy Efficacy trial [CREATE])58 and mirtazapine (Myocardial Infarction and Depression Intervention Trial [MIND-IT])59 have been evaluated in clinical trials involving patients with CHD.

SADHART studied depression after ACS over 6 months. Depression scores in patients taking sertraline improved significantly more than in those receiving placebo. Most patients were also prescribed aspirin, statins and β-blockers. Life-threatening cardiovascular events occurred less frequently in the sertraline group; however, this result was not statistically significant.56

ENRICHD was a large trial that evaluated the effect of cognitive behaviour therapy (CBT) on depression or low social support in patients with a recent MI. Depression was diagnosed in 74% of participants. CBT improved depression but failed to reduce the number of CHD events. Patients whose depression did not respond to CBT were referred for treatment with antidepressant drugs. Selective serotonin reuptake inhibitors (SSRIs) (mainly sertraline) significantly improved depression in those patients. In the SSRI-treated group, there was a 43% (P < 0.005) reduction in deaths or recurrent MIs.57 However, this was a subset analysis, and therefore is hypothesis generating only.

In the Canadian CREATE trial58 and the MIND-IT trial,59 there were too few CHD events reported to enable analysis of cardiovascular outcomes.

Tricyclic antidepressants may worsen CHD outcomes and should be avoided in patients with CHD. Importantly, tricyclic antidepressants have been associated with increased mortality in patients with CHD.60–62 In contrast, a recent meta-analysis of trials involving SSRIs in patients with CHD concluded that this class of drugs was well tolerated, with the risk of adverse events being similar to that for placebo.63

Psychological therapy

Of the various psychological therapies, CBT and integrative therapies (eg, interpersonal psychotherapy) have the best documented efficacy for treatment of major depressive disorder.64,65 CBT was used in the ENRICHD trial,57 interpersonal psychotherapy in the CREATE trial,58 and problem-solving therapy in the COPES trial.52,53 These therapies were all beneficial for depression but did not affect CHD outcomes. The efficacy of psychological therapy as a treatment for major or minor depression was evaluated in patients who underwent coronary artery bypass surgery.66 Significantly more patients in the CBT group (71%) and the stress management group (57%) had low depressive symptom levels, compared with those having usual care (33%), and these results were maintained at 6 months.66

A Cochrane review of psychological interventions for patients with CHD found evidence of small-to-moderate improvements in depression and anxiety symptoms with such interventions, but no strong evidence that the interventions reduced total deaths, risk of revascularisation, or non-fatal infarction.67 The interventions that were less effective were those that aimed to educate patients about cardiac risk factors; those that included client-led discussion and emotional support; or those that included family members in the treatment process.67 Uncertainty remains regarding the subgroups of patients who would benefit most from psychological treatments and the characteristics of successful interventions.

Exercise

Many patients with mild depression respond well to regular exercise and cardiac rehabilitation (exercise-based). A recent Cochrane review of exercise as a treatment for depression concluded that exercise improves depression with a similar efficacy to CBT.68 The benefit of exercise appears to have a dose–response relationship, needing at least 30 minutes of moderate aerobic activity on 5 days per week.69,70 This is consistent with usual public health recommendations.

The benefit of exercise in patients with CHD and depression has been demonstrated in the recent UPBEAT (Understanding the Prognostic Benefits of Exercise and Antidepressant Therapy) trial.71 Patients with CHD with at least a mildly elevated score for depressive symptoms (BDI score > 7) were allocated at random to treatment with SSRIs, exercise or neither. Exercise was equivalent to SSRI treatment in improving depression scores, with patients in both groups showing greater improvement than the control group.71 In a large randomised controlled trial (RCT) of 2322 patients with heart failure (of whom 28% were depressed), exercise, in addition to reducing mortality and hospitalisation (P = 0.03), significantly reduced depression (P = 0.002).72

Complementary and alternative therapies

Up to 50% of patients with depression have been shown to use complementary and alternative medicines without disclosing this to their treating clinician.73 Therapies that may be effective in depression are supplemental marine n-3 fatty acids (eicosapentaenoic acid [EPA] and docosahexaenoic acid [DHA]), S-adenosylmethionine (SAMe) and St John’s wort.74 Specific trials in patients with CHD and depression have not been performed with the latter two therapies.

Marine n-3 fatty acids (at a dose of 1 g per day of combined EPA–DHA) are recommended by the NHFA and the AHA for all patients with CHD.75 This dose also may improve mild depression. However, the addition of 2 g/day of combined EPA and DHA to sertraline 50 mg daily for depressive symptoms appears to provide no added benefit over sertraline alone.76 Some trials comparing St John’s wort and SAMe to antidepressant medications suggest a similar effectiveness in improving depression to antidepressant medications.77–79 However, most commercial brands of St John’s wort have not undergone randomised trials.78,79

Adherence

Depression is a major predictor of poor adherence in patients with CHD, be it to drug therapy or lifestyle measures.80,81 Patients with depression are three times more likely to be non-compliant with medical treatment than patients without depression.82 Greater severity and chronicity of depression have been associated with poorer adherence to aspirin therapy after MI.83 Adherence to aspirin therapy after ACS has been shown to be significantly lower in persistently depressed patients (76.1%) than in those whose depression improved (87.4%), or who were not depressed (89.5%).83 Patients who are persistently depressed are also less likely to undertake behaviours that reduce risk; for example, quitting smoking, taking medications, exercising and attending cardiac rehabilitation.81 The SADHART trial showed adherence to medication increased after remission of depression in 68.4% of participants taking the trial medication.84

A recent RCT of a collaborative-care depression treatment program in 134 patients with depression after ACS demonstrated improved adherence to medications and secondary prevention behaviours and was independently associated with improvement in depression.85 However, in another RCT of 157 patients undergoing treatment for depression after ACS, there were no improvements in adherence to risk-reducing behaviours in spite of a significant reduction in depression.86

Referral

Once depression is identified through screening, treatment may be initiated immediately, or referral to psychological or psychiatric services may also be considered appropriate. Most patients with depression in Australia are managed by general practitioners.87

Members of the Cardiac Society of Australia and New Zealand, the majority of whom are clinical cardiologists, were surveyed regarding assessment of depression. Most respondents screened for depression occasionally, with only 3% using a formal tool. Lack of confidence in identifying depression was the strongest predictor of a low screening frequency. Cardiologists rarely initiated treatment for depression, and 43% did not feel they were responsible for treating depression.88

There can be a reluctance to treat depression in patients with CHD because of a belief that depression is normal after an acute cardiovascular event. Mild depression may resolve spontaneously; however, for most individuals with CHD, depression remains long term.89

Conclusion

A summary of the key evidence-based points is provided in Box 3 and Box 4, and the Appendix gives the National Health and Medical Research Council grades of recommendations and evidence hierarchy.

High-quality care for treatment of depression is achievable and affordable. The benefits of treating depression in people with CHD include improved quality of life, improved adherence to other therapies and potentially improved CHD outcomes.90 Effective treatment of depression may decrease CHD events but this is not proven, as no adequately powered trials have been completed, nor are there any ongoing.

1 Process used to develop this National Heart Foundation of Australia consensus statement

The Expert Working Group members performed relevant literature searches using key search phrases including, but not limited to, “stress”, ”depression”, “anxiety”, “treatment of depression”, “acute coronary syndromes”, “adherence and depression” and “screening for depression”. This was complemented by reference lists compiled from reviews and personal collections of the Expert Writing Group members.

Searches were limited to evidence available for human subjects with coronary heart disease published in English up to December 2012. The recommendations made in this consensus statement have been graded according to the National Health and Medical Research Council guidelines (see Appendix).2 The Cardiac Society of Australia and New Zealand, beyondblue: the national depression initiative and the Royal Australian and New Zealand College of Psychiatrists were consulted during the development of this document and have endorsed its content.

2 Patient Health Questionnaire (PHQ-2) Yes/No version35

During the past month, have you often been bothered by feeling down, depressed or hopeless?
During the past month, have you often been bothered by little interest or pleasure in doing things?

3 National Heart Foundation of Australia grades of recommendation and levels of evidence for screening, referral and treatment for depression in patients with coronary heart disease (CHD)2

Recommendation

Grade2

Level2

For patients with CHD, it is reasonable to screen for depression

Treatment of depression in patients with CHD is effective in decreasing depression

Treatment of depression in patients with CHD improves CHD outcomes

Treatment of depression in patients with CHD changes behavioural risk factors/adherence

III-2

Exercise is an effective treatment of depression in patients with CHD

Exercise improves CHD outcomes in patients with CHD

Psychological interventions improve depression in patients with CHD

Psychological interventions improve CHD outcomes in patients with CHD and depression

SSRIs improve depression in patients with CHD

SSRIs improve CHD outcomes in patients with CHD and depression

III-1

Collaborative-care approach improves depression in patients with CHD

Collaborative-care approach improves CHD outcomes in patients with CHD and depression

SSRIs = selective serotonin reuptake inhibitors.

4 Treatment of depression in patients with coronary heart disease (CHD) — summary of treatment subgroup effects showing grade of recommendation and level of evidence

Treatment

Depression

CHD outcome

Grade2

Level2

Grade2

Level2

Non-drug

Exercise

Psychological, including CBT

St John’s wort*

—*

n-3 fatty acids

D†

SAMe*

—*

Collaborative

Drug

SSRIs

III-1

CBT = cognitive behaviour therapy. SAMe = S-adenosylmethionine. SSRIs = selective serotonin reuptake inhibitors. * Insufficient evidence to rate or no trials have been performed. † Data not available in patients with CHD.

Appendix: Definition of National Health and Medical Research Council (NHMRC) grades of recommendations and evidence hierarchy*

Definition of NHMRC grades of recommendations

Grade

Description

Body of evidence can be trusted to guide practice

Body of evidence can be trusted to guide practice in most situations

Body of evidence provides some support for recommendation(s) but care should be taken in its application

Body of evidence is weak and recommendation must be applied with caution

NHMRC evidence hierarchy: designation of levels of evidence

Level

Intervention

A systematic review of level II studies

A randomised controlled trial

III-1

A pseudorandomised controlled trial (ie, alternate allocation or some other method)

III-2

A comparative study with concurrent controls:

Non-randomised, experimental trial
Cohort study
Case—control study
Interrupted time series with a control group

III-3

A comparative study without concurrent controls:

Historical control study
Two or more single arm study
Interrupted time series without a parallel control group

Case series with either post-test or pre-test/post-test outcomes

* From NHMRC additional levels of evidence and grades for recommendations for developers of guidelines.2

The use, misuse and abuse of dabigatran

A number of issues need to be considered before using dabigatran routinely in clinical practice

There has been much publicity and controversy about the role of dabigatran in the prevention of embolic stroke in those with atrial fibrillation since the landmark RE-LY (Randomized Evaluation of Long-Term Anticoagulation Therapy) trial was published in September 2009.1 In this perspective, we provide a critical appraisal of the original trial and highlight some concerns (“use”), review some of the issues about bleeding and inappropriate selection of patients (“misuse”), and detail some of the heated and unique politics around its attempted listing on the Pharmaceutical Benefits Scheme (PBS) in Australia (“abuse”).

Use of dabigatran

A rigorous critical appraisal of the RE-LY trial reveals a few concerns.

Selection bias: Just under a third of participants enrolled
in the trial had a CHADS2 (congestive heart failure, hypertension, age ≥ 75 years, diabetes, 1 point each; prior stroke or transient ischaemic attack [TIA], 2 points) score of 0 or 1. These people have a low risk of embolism (0.5%–1.7% per year)2 and are a group where the use of anticoagulation is considered as an option, and an “individualised” approach is recommended.3 Including low-risk participants for whom warfarin may be of marginal net benefit but who still carry the bleeding risk is likely to bias results in favour of dabigatran; this affects the internal validity of the trial, not just the generalisability.

Effect of co-interventions: About 40% of participants in the trial were concurrently taking aspirin and half of these (20%) continued to take aspirin throughout the trial. Given the evidence that the combination of aspirin and warfarin increases the risk of bleeding and is only recommended in the setting of prosthetic valves,4 this would potentially have increased the risk of bleeding in the warfarin group. Although limited evidence would suggest that this risk may not be differential between warfarin and dabigatran,5 the rate of major bleeding in the RE-LY trial (over 3% per year) is higher than the 1%–1.5% per year usually expected.

Adverse events: The original trial showed a significantly increased risk of myocardial infarction (MI) in the dabigatran group with a relative risk of 1.38 (95% CI, 1.0–1.9, P = 0.048) compared with warfarin. Although it was noted that this was increased, it was described as being of uncertain significance. Revised outcomes were published for the RE-LY trial and included 28 previously unreported silent MIs, which tipped the balance away from the dabigatran group and reduced the relative risk to 1.27 (95% CI, 0.94–1.71, P = 0.12). Nevertheless, a recent meta-analysis pooling seven trials of dabigatran found a statistically increased risk of MI or acute coronary syndrome, with a pooled odds ratio of 1.33 (95% CI, 1.03–1.71, P = 0.03), and this remained significant even with the revised outcome figures.6 It is important to note that the increased risk of MI was seen even with 20%–40% of people in the RE-LY trial taking aspirin; the rate might have been even higher if participants had not been taking aspirin. In practice, however, physicians do not coprescribe dabigatran and aspirin, and there is an assumption that dabigatran
will also provide equivalent protection for coronary artery disease, which is not supported by the data. There was also a significantly increased risk of gastrointestinal haemorrhage. Neither of these increased risks (MI and gastrointestinal bleeding) was included in the marketing summary that was sent out as part of the patient familiarisation program (PFP).

Generalisability: The population in the RE-LY trial overrepresented those with a previous stroke or TIA (about 20%) and underrepresented those with a previous MI (about 17%). In a typical older community cohort, one would expect prior MI to be about twice as prevalent as prior stroke or TIA. In generalising the results of the trial to a broader population, one might expect the absolute increase in cardiac risk with dabigatran to be even higher.

Conclusion: These issues all feed into a bias in favour of dabigatran, causing its benefit to be overstated when generalising to a broader population.

Misuse of dabigatran

Examples have already been published of patients who were in a stable condition while taking warfarin and who were switched to dabigatran with ensuing major bleeds.7 The United States Food and Drug Administration is also evaluating postmarketing reports of serious bleeding events in patients taking dabigatran. A number of issues need to be considered in deciding who is appropriate for dabigatran.

Stability on warfarin: Patients with a stable condition while taking warfarin, with no adverse events, should probably continue taking it. Two separate economic analyses show that where the international normalised ratio (INR) is therapeutic over 65%–70% of the time, the incremental cost-effectiveness ratio increases, making warfarin more cost-effective than dabigatran.8,9

Risk of bleeding: The effects of dabigatran cannot be pharmacologically reversed, but given that it is about one-third protein bound, it can be dialysed in patients with renal impairment, with about 60% removed after 2–3 hours of dialysis. Work on an antidote is in process. Those who bleed while taking dabigatran are usually supported until the drug clears sufficiently from the system, unlike rivaroxaban or warfarin, which can be reversed with prothrombin complex concentrate or vitamin K, respectively.10

Risk of cardiac disease: As stated earlier, dabigatran does not provide the same degree of cardioprotection as warfarin, and the underlying risk of cardiac disease must be considered. It may be the first agent that clinicians use that does not have a similar protective effect against cardiac and neurological risk. Those with moderate to high underlying cardiac risk must be considered for concomitant aspirin, although this must be weighed with the higher risk of bleeding, particularly gastrointestinal bleeding, with dabigatran.

Elderly patients: Dabigatran is renally excreted and must
be used with caution in those with impaired renal function (estimated glomerular filtration rate [eGFR], 30–50 mL/min) and is contraindicated in those with renal failure (eGFR, < 30 mL/min). Elderly patients also have changes in lean body mass that are not captured in the eGFR.

Interacting drugs: Although dabigatran has a wide therapeutic range, it is a prodrug that is metabolised by the
P-glycoprotein transporter, and hence drugs that interact with P-glycoprotein can increase dabigatran concentrations (eg, macrolides, amiodarone and verapamil) or reduce them (eg, rifampicin, carbamazepine and phenytoin). These considerations are especially important in the elderly.

Conclusion: Issues such as stability on warfarin, high risk
of bleeding or cardiovascular disease, altered renal function and interactions with other drugs should be weighed when considering prescription of dabigatran, especially in the elderly.

Abuse of dabigatran

The potential market and profits for dabigatran are large; this provides considerable incentive to aggressively market a new product and has led to some heated politics, perhaps more extreme than any seen to date. We describe the politics and marketing of dabigatran in Australia as an example.

Listing on the PBS: In Australia alone, the Pharmaceutical Benefits Advisory Committee (PBAC) predicted costs to the PBS of over $100 million by the 5th year after the introduction of dabigatran, although there would be some savings with a reduction in INR testing. The PBAC recommended the listing of dabigatran but noted: “The listing of dabigatran may also result in patients at low risk currently managed on aspirin or no treatment being unnecessarily transferred to dabigatran at a much higher cost.”11 Given the potentially significant financial impact of listing dabigatran and concerns about its place in the treatment of atrial fibrillation, the Minister for Health recommended a review of anticoagulant therapies in atrial fibrillation before any listing, with broad terms of reference that incorporated minimising harm, improving health outcomes, modifying health systems delivery and monitoring cost. This generated unprecedented criticism of the government.12 The proposed method of listing is possibly the cause of this criticism. The listing as “streamlined authority” means no phone call to regulating authorities is required. The PBAC itself admitted an inability to control prescribing: “Medicare Australia would not be able to enforce compliance with the risk factors under the requested ‘streamlined’ authority.”11

Familiarisation programs: A PFP is designed with “the aim of allowing the medical profession to evaluate and become familiar with the product”.13 Normally the medication would be limited to specialists in a particular field, with dispensing at hospital pharmacies. In the case of dabigatran, Boehringer Ingelheim launched an extensive outreach to general practitioners, shouldering the cost of dabigatran for 10 patients per GP until such time as the PBS approves the listing. In addition, copies of promotional material were sent to every registered medical practitioner in Australia, including hospital residents and advanced trainees. This went far beyond the usual scope of a PFP and could not be considered in keeping with the spirit of a PFP, but it did not formally breach the Medicines Australia guidelines.

Web campaign: Boehringer Ingelheim launched a website called “vote against stroke”, which encouraged people to write to their member of parliament to protest the delay in approving dabigatran. This was seen as unduly coercive and was heavily criticised by the media and public, forcing the company to shut down the website.14

Conclusion: The marketing of dabigatran exemplifies a much broader marketing strategy than that usually seen in PFPs, and requires GPs to be much more informed about the evidence base.

Summary

The tale of dabigatran sounds some cautionary notes about proper critical appraisal of new randomised controlled trials, care in deciding on the generalisability of results, judicious screening of patients and lessons about the politics around increasingly lucrative drugs. The old lesson of caveat utilitor still holds: let the user beware!

Is there really misuse and abuse of dabigatran?

A brief commentary on the RE-LY study

Appropriate and optimal anticoagulation for the increasing number of Australians with non-valvular atrial fibrillation (AF) remains a challenge. If achieved, it would substantially reduce the burden of disabling and fatal AF-related stroke.

Four large randomised controlled trials (RCTs) have reported that the direct thrombin inhibitor, dabigatran etexilate, and the direct activated factor X (Xa) inhibitors, rivaroxaban and apixaban, are at least as efficacious and safe as warfarin, and apixaban is superior to aspirin, in a broad range of individuals with non-valvular AF.1–5 On the basis of these results, regulators have approved dabigatran in more than 75 countries, including Australia, and guidelines suggest that the new oral anticoagulants are preferable to warfarin for most patients with non-valvular AF.6–9

Patient selection in the RE-LY trial

The RE-LY (Randomized Evaluation of Long-Term Anticoagulant Therapy) trial of dabigatran versus warfarin included patients with all degrees of risk of stroke and systemic embolism.9 If the trial had excluded participants with a low risk, ie, a CHADS2 score of 1 (congestive heart failure, hypertension, age ≥ 75 years, diabetes, 1 point each; prior stroke or transient ischaemic attack, 2 points), clinicians would not now know that these patients derive benefit from dabigatran compared with warfarin.

Quality of anticoagulation control

RE-LY trial participants who were assigned warfarin had a median time in therapeutic range (TTR) of 67%. This may have led to an underestimate of the benefits of dabigatran compared with warfarin in community practice, where the median TTR for patients taking warfarin is rarely this high.

Benefits of dabigatran

Dabigatran (150 mg twice daily) compared with warfarin produced reductions in ischaemic stroke (one-third)
and haemorrhagic stroke (one-half) that were statistically significant and clinically important. The reduction in stroke with dabigatran was also evident in the very elderly, participants with renal impairment, participants with previous myocardial infarction (MI) or stroke, and in the presence and absence of aspirin.

Safety of dabigatran

Both doses (110 mg or 150 mg twice a day) of dabigatran were associated with significantly lower life-threatening and fatal bleeding than warfarin. Among participants aged ≥ 75 years, extracranial bleeding risk was similar or higher with both doses of dabigatran compared with warfarin, whereas intracranial bleeding risk was lower with both doses of dabigatran.10 In participants aged < 75 years, both doses of dabigatran were associated with lower risks of both intracranial and extracranial bleeding than warfarin.

Participants who were taking dabigatran also had lower rates of bleeding if they required urgent or emergency surgery than those taking warfarin, despite access
to vitamin K and clotting factors that could pharmacologically reverse the anticoagulant effect
of warfarin.11 Although there is no specific antidote
for dabigatran, the case-fatality rate of intracranial haemorrhage was not significantly different in participants assigned dabigatran or warfarin.12

The absolute increase in MI with dabigatran compared with warfarin was 0.14%–0.17% per year. This was outweighed by a 0.6% per year reduction in stroke and systemic embolism.13 Adding aspirin to dabigatran did not protect against any increased risk of MI, but increased bleeding.

Real world experience with dabigatran

Recent postmarketing surveillance by the European Medicines Agency and the United States Food and Drug Administration (FDA) found no evidence of excess serious bleeding with dabigatran compared with warfarin.14,15

We have a continued responsibility to ensure that the favourable effects of new anticoagulants compared with warfarin demonstrated in RCTs are translated into clinical practice. Ongoing Phase IV safety surveillance, economic analysis, and doctor and patient education are key to this process.

Meanwhile, optimal safety and effectiveness of the new oral anticoagulants in routine clinical practice demand appropriate selection of patients (eg, estimated glomerular filtration rate ≥ 30 mL/min) and dose, a high level of adherence, regular monitoring of renal function (which may deteriorate in at-risk patients) and appropriate periprocedural management of anticoagulation interruption for invasive procedures. As treatment with
a new oral anticoagulant is not suitable for all patients, warfarin will retain an important role in the management of patients who can maintain excellent anticoagulant control, as well as those with severe renal impairment and other contraindications to the new oral anticoagulants.

The role of depression in the primary prevention of cardiovascular disease

To the Editor: We would like to add a cautionary note to O’Neil’s support for the inclusion of “depression and other psychosocial factors” in the updated National Vascular Disease Prevention
Alliance guidelines for assessing cardiovascular disease (CVD) risk. Her statement “it is now clear that depression is also an important risk factor for CVD” is premature.1 Different authors have challenged the suggested relationship and, given divergent findings and opinions, it is misleading to claim that the matter is now “clear”.2–5 We do not deny a possible link between depression and coronary heart disease (CHD) but suggest that the extent and nature of the relationship has yet to be clarified. Premature acceptance and promotion of the idea that depression is a risk factor for CHD might contribute to overdiagnosis and overtreatment of depression as well as undue worry about the risk of CHD by individuals diagnosed with depression.

The role of depression in the primary prevention of cardiovascular disease

In reply: I thank Stampfer, Hince and Dimmitt for their response.
While I acknowledge that there are aspects of the relationship between cardiovascular disease (CVD)
and depression that remain undetermined, recent evidence clearly supports the role of depression as an independent risk factor for CVD and is indeed convincing.1–3 This is especially true in studies that assess depression using diagnostic criteria.4 We know that the risk of developing coronary heart disease (CHD) for individuals with depression is twofold, and these individuals have a similar risk of CVD-related death.2 We also know that the independent contribution of depression to CVD is at least comparable to that of more traditional risk factors including diabetes, hypercholesterolaemia, smoking or obesity.5 A decade has now passed since the National Heart Foundation’s seminal position paper concluded that: “there is strong
and consistent evidence of an independent causal association between depression, social isolation and lack of quality social support and the causes and prognosis of CHD”.5 Despite this, depression remains underestimated as a meaningful contributor to CVD.