A systematic review update

Transcription

1 A systematic review update May 2009 Systematic review update: 64-slice or higher computed tomography in the investigation of patients with suspected coronary artery disease Dan Paech Adele Weston

2 This report should be referenced as follows: Paech D and Weston A. Systematic review update: 64-slice or higher computed tomography in the investigation of patients with suspected coronary artery disease HSAC Report 2009; 2(12) Health Services Assessment Collaboration (HSAC), University of Canterbury ISBN (online) ISBN (print) ISSN (online) ISSN X (print)

3 i Review Team The evidence review will be undertaken by the Health Services Assessment Collaboration (HSAC). HSAC is a collaboration of the Health Sciences Centre of the University of Canterbury, New Zealand and Health Technology Analysts, Sydney, Australia. This report was authored by Dan Paech, Health Outcomes Analyst, who developed and undertook the literature search, extracted the data, conducted the critical appraisals, and prepared the clinical component of the report, and Dr Adele Weston who provided guidance and reviewed report. The primary reviewer contact for this project is Dr Adele Weston. Acknowledgements Dr Ray Kirk and Dr Adele Weston (as HSAC Directors) reviewed the final draft. Cecilia Tolan (Administrator) provided administrative support. The current review was conducted under the auspices of a contract funded by the New Zealand Ministry of Health. This report was requested by Sandy Dawson, Chief Advisor, The Sector Capability and Innovation Directorate of New Zealand s Ministry of Health. We thank Sandy Dawson for his assistance in developing the scope of the review and providing clinical practice information for the review. Technical information was provided by Dr Niels van Pelt. The systematic review of the evidence will ultimately be used by The Sector Capability and Innovation Directorate team to inform policy decision making in conjunction with other information. The content of the review alone does not constitute clinical advice or policy recommendations. Copyright Statement & Disclaimer This report is copyright. Apart from any use as permitted under the Copyright Act 1994, no part may be reproduced by any process without written permission from HSAC. Requests and inquiries concerning reproduction and rights should be directed to the Director, Health Services Assessment Collaboration, Health Sciences Centre, University of Canterbury, Private Bag 4800, Christchurch, New Zealand HSAC takes great care to ensure the accuracy of the information in this report, but neither HSAC, the University of Canterbury, Health Technology Analysts Pty Ltd nor the Ministry of Health make any representations or warranties in respect of the accuracy or quality of the information, or accept responsibility for the accuracy, correctness, completeness or use of this report. The reader should always consult the original database from which each abstract is derived along with the original articles before making decisions based on a document or abstract. All responsibility for action based on any information in this report rests with the reader. This report is not intended to be used as personal health advice. People seeking individual medical advice should contact their physician or health professional.

4 ii The views expressed in this report are those of HSAC and do not necessarily represent those of the University of Canterbury New Zealand, Health Technology Analysts Pty Ltd, Australia or the Ministry of Health. Contact Details Health Services Assessment Collaboration (HSAC) Health Sciences Centre University of Canterbury Private Bag 4800 Christchurch 8140 New Zealand Tel: Fax: hsac@canterbury.ac.nz Web Site: healthsac.net

5 iii Executive Summary Introduction This systematic review update aimed to provide a summary of the recent evidence pertaining to the clinical effectiveness of 64-slice or higher computed tomography (CTA) as an alternative to invasive coronary (ICA) in the investigation of patients with suspected coronary artery disease (CAD). ICA allows ad hoc performance of coronary interventions such as percutaneous coronary intervention (PCI) and coronary artery bypass graft (CABG), however only one third of all ICA s in Europe are performed in conjunction with a revascularisation procedure, whilst the rest are performed only for diagnostic purposes (Togni et al 2004). Therefore if CTA proves to be a successful diagnostic performance measure, it could prevent the application of some invasive diagnostic procedures (Piers et al 2008). Methods This systematic review update was based on a health technology assessment performed in the United Kingdom (UK) by Mowatt and colleagues (2008). A systematic method of literature searching and selection was employed in the preparation of this review update with searches limited to material published from December 2006 onwards. The search identified 1438 citations and after applying study selection criteria 28 studies were included for review. The included studies were quality assessed using the NHMRC diagnostic levels of evidence and a modified version of the QUADAS tool. Data were extracted onto specifically designed data extraction forms and a summary of the study characteristics and calculated diagnostic performance (i.e. sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and overall diagnostic accuracy) are provided in the results section. Results were also meta-analysed and pooled sensitivity, specificity, PPV, NPV and diagnostic accuracy results are presented at the patient, vessel and segment level. Key results In general, the pooled results of this meta-analysis update demonstrate the high diagnostic accuracy of 64-slice CTA in patients with suspected CAD. Across the included trials there was remarkable consistency in the sensitivity and NPV, considered the most important measures for this technology within this context. The base case meta-analysis (i.e. studies with equivocal test results omitted excluded) at the patient-level, indicated a sensitivity of 98.2%, specificity of 81.6%, PPV of 88.9%, NPV of 96.8%, and diagnostic accuracy of 91.6%. Pooled diagnostic performance results at the vessel and segment level supported the patient-level findings. In all vessels, the pooled sensitivity was 95.0%, specificity 85.2%, PPV 69.4%, NPV 97.9%, and diagnostic accuracy 87.7%. At the individual artery level, overall diagnostic accuracy appeared to be slightly higher in the left and right coronary artery and slightly lower in the left anterior descending and circumflex

6 iv artery. In all segments, the sensitivity was 91.1%, specificity 94.3%, PPV 65.7%, NPV 98.9%, and overall diagnostic accuracy 94.0%. Conclusions There was substantial evidence showing the high diagnostic accuracy of 64-slice CT in patients with suspected CAD. The high sensitivity observed in this update indicates that CTA can effectively identify the majority of patients with significant stenosis ( 50%). The high NPV at the patient, vessel and segment level establishes CTA as a highly effective non-invasive alternative to ICA for the exclusion of significant coronary artery stenosis. The extent to which CTA is cost-effective in the NZ DHB setting is yet to be determined.

7 v Table of Contents Review Team... i Acknowledgements... i Copyright Statement & Disclaimer... i Contact Details... ii Executive Summary... iii Introduction... iii Methods... iii Key results... iii Conclusions... iv Table of Contents... v List of Tables... vii List of Figures... viii List of Abbreviations and Acronyms... ix Introduction... 1 Objective... 1 Coronary artery disease... 1 Computed tomography... 2 Invasive coronary... 2 Clinical management algorithm... 3 Structure of report... 3 Methods... 5 Research questions... 5 Level of analysis... 6 Literature search... 7 Assessment of study eligibility Appraisal of included studies Dimensions of evidence Data extraction Data synthesis Limitations of the review methodology Results Overview Included study characteristics Diagnostic accuracy of CTA Meta-analysis of diagnostic performance measures... 36

8 vi Patient-level analysis Vessel and segment level analysis Summary and Conclusions Summary of evidence for evidence review Limitations of evidence base Conclusions References Appendix A: Included Studies Appendix B: Excluded Studies Annotated by Reason for Exclusion Appendix C: Study Characteristics Appendix D: Study Quality (QUADAS) Appendix E: Observed Diagnostic Results

9 vii List of Tables Table 1: Recent systematic review, selected for updating... 5 Table 2: Criteria for determining study eligibility... 6 Table 3: Nature of the evidence... 6 Table 4: Search strategy... 8 Table 5: Application of selection criteria to citations Table 6: Dimensions of evidence (NHMRC, 2000) Table 7: NHMRC diagnostic levels of evidence Table 8: Modified QUADAS quality assessment checklist (derived from QUADAS tool) Table 9: Reporting biases in systematic reviews * Table 10: Summary of included study characteristics Table 11: Diagnostic performance of CTA in the included studies Table 12: Pooled CTA diagnostic performance measures Table 13: Pooled diagnostic performance measures: studies with equivocal test results omitted are excluded... 40

10 viii List of Figures Figure 1: Figure 2: Figure 3: Figure 4: Figure 5: Figure 6: Initial clinical management algorithm for the management of patients with suspected CAD... 3 Scatter plot showing the correlation between the prevalence of CAD and PPV in studies included in the base case patient level analysis Forest plots: sensitivity and specificity for CTA in detecting significant stenosis at the patient-level (base-case) Forest plots: PPV and NPV for CTA in detecting significant stenosis at the patient-level (base-case) Forest plots: sensitivity and specificity for CTA in detecting significant stenosis at the patient-level (alternative analysis) Forest plots: PPV and NPV for CTA in detecting significant stenosis at the patient-level (alternative analysis)... 39

11 ix List of Abbreviations and Acronyms AIHW Australian Institute of Health and Welfare BPM Beats per minute CABG Coronary artery bypass graft CAD Coronary artery disease CNC Could not calculate CT Computed tomography CTA Computed tomography CX Circumflex ECG Electrocardiogram HR Heart rate HTA Health technology assessment ICA Invasive coronary LAD Left anterior descending LM Left main MPS Myocardial perfusion scintigraphy MSAC Medical Services Advisory Committee MSCT Multi-slice computed tomography NHMRC National Health and Medical Research Council NHS National Health System NICE National Institute of Clinical Excellence NPV Negative predictive value NR Not reported PCI Percutaneous coronary intervention PPV Positive predictive value QUADAS Quality Assessment of Diagnostic Accuracy Studies RCA Right coronary artery SD Standard deviation UK United Kingdom VCT Volume computed tomography

12

13 1 Introduction Objective The purpose of this systematic review update is to provide a summary of the recent evidence pertaining to the relative clinical effectiveness of 64-slice or higher computed tomography (CTA) as an alternative to invasive coronary (ICA) in the investigation of patients with suspected coronary artery disease (CAD). Specifically, evidence subsequent to December 2006 (i.e. the last search date of the systematic review chosen for updating), forms the basis of the report. The review has been requested in order to improve the management of cardiovascular health amongst New Zealanders. Reducing the incidence and impact of cardiovascular disease is a key national health priority in New Zealand, with cardiovascular risk screening and management one area of particular focus. The CTA technology is an option for consideration by the Quality Improvement Plan for Cardiovascular Disease and Diabetes. Coronary artery disease Cardiovascular disease is the leading cause of death in Australia and New Zealand, accounting for 35% of all deaths in Australia in 2005, and 40% of all deaths in New Zealand in 2000 (AIHW 2008 and Hay 2004). A substantial portion of this burden is attributable to CAD, which is the largest single cause of mortality accounting for 18% of all deaths in Australia in 2005 and 22% in New Zealand in 2000 (AIHW 2008 and Hay 2004). CAD presents itself in many forms including angina, myocardial infarction and sudden death. It is caused by the process of atherosclerosis, which compromises coronary blood supply. Angina occurs when there is incomplete narrowing of the coronary artery, whereas acute myocardial infarctions are caused by abrupt occlusion of one or more coronary arteries, which may be fatal (MSAC 2003). Data from the New Zealand Health Survey conducted in 2002/03, indicated that one in 10 adults (10.4%) had been previously diagnosed with heart rate disease (i.e. heart attack, angina, abnormal heart rhythm or heart failure). There was no significant difference in the prevalence of heart disease in males (9.6%) and females (8.4%). In both males and females, the prevalence was higher in Maori than in European or Asian and Pacific ethnic groups. There was also a significantly higher prevalence of heart disease in individuals over 55 years of age with the highest prevalence in the 75 plus age group (Ministry of Health 2004). In , admissions to public hospitals for coronary heart disease totalled 29,456, 59% men and 41% women (Hay 2004). Based on data from the National Health Survey, approximately 3.2% of Australians have CAD, corresponding to around 637,900 people. Among those with CAD, around 359,500 had angina and 354,700 people had had a heart attack (note that a person may have reported more than one condition) (Australia s Health 2008). The prevalence of angina was higher among males than females, at 2.2% and 1.4% respectively. For a history of heart attack, the difference between males and females

14 2 was greater still, with prevalence s of 2.6% and 1.0% respectively. CAD predominantly affects middle-aged and older Australians, with the majority of hospital admissions for heart attack and cardiac procedures occurring among the population aged 60 years and over 70% of acute myocardial infarction hospital admissions, 73% of coronary artery bypass graft (CABG) procedures and 61% of percutaneous coronary intervention (PCI) procedures (AIHW 2008). A large part of the death, disability and illness caused by CAD is preventable. The main behavioural risk factors for CAD are a high-fat, excess energy diet typical in Western countries, physical inactivity and cigarette smoking (AIHW 2002). Computed tomography Computed tomography can be used to visualise narrowing of the coronary arteries and diagnose coronary heart disease. It combines the use of X-rays with advanced computerised analysis. CTA scanners have multiple rows of detectors, enabling a large number of thin image slices (ranging from 4 to 256) to be obtained simultaneously. CT technology has advanced rapidly with four-slice machines appearing in 1998, 16-slice in 2001, 64-slice machines at the end of 2004 and 256- slice machines in 2006 (Mowatt et al 2008). There has been considerable enhancement in temporal and spatial resolution which has been reported to result in clinical benefit in terms of improved diagnostic accuracy (Brodoefel et al 2008). The newer multi-detector machines can produce more images in less time, thereby increasing throughput and theoretically decreasing the cost per patient. At present, CTA is mainly used for detecting or excluding significant coronary stenosis ( 50% diameter reduction) in coronary arteries. CTA is less invasive, carries a lower risk of serious complications and is cheaper than standard invasive coronary (ICA) (NICE 2007). It is important to note that CTA is not proposed for use in asymptomatic presenting patients (i.e. population screening). As the clinical management algorithm shows in Figure 1, CTA is primarily used in patients with suspected CAD who have already undergone an inconclusive exercise treadmill test. Invasive coronary Invasive coronary (ICA), also referred to as conventional, was first developed in the late 1950 s as a minimally invasive technique to visualise the coronary arteries, and hence assess and diagnose CAD (Ryan 2002). Typically, ICA involves inserting a catheter though the femoral, brachial or radial artery and threading it back to reach the heart and coronary arteries (Popma and Bittl 2001). Due to its invasive nature, it has been reported that ICA is not suitable as a general screening procedure thus prompting interest in the development of non-invasive coronary imaging (Wittlinger et al 2002). However, ICA remains the diagnostic criterion standard for clinical evaluation of known or suspected CAD owing to its high temporal and spatial resolution. It is conducted both to assist with determining the extent of CAD and to help guide cardiac interventions to treat these disorders.

15 3 Clinical management algorithm When patients present with suspected CAD, a resting electrocardiogram (ECG) is normally performed in the first instance. If this is normal or equivocal, an exercise stress test will usually follow. Other non-invasive techniques have been developed to assist in the risk assessment process including CTA and myocardial perfusion scintigraphy (MPS). Although current clinical practice varies, one clinical management algorithm for the initial investigation of suspected CAD is shown below in Figure 1. Figure 1: Initial clinical management algorithm for the management of patients with suspected CAD Confirm CAD a Go to assessment pathway Symptoms CAD Resting ECG Positive Go to assessment pathway Normal or equivocal Exercise test b Equivocal c Negative High chance CAD Further test MPS CTA Not possible d Further test Low chance CAD CTA Reassure and discharge MPS Source: Mowatt et al (2008), Figure 1, page 17 Abbreviations: CAD = coronary artery disease; CTA = computed tomography ; ECG = electrocardiogram; MPS = myocardial perfusion scintigraphy a for example, evidence of previous myocardial infarction b exercise test plus clinical finding c includes unable to exercise enough d e.g. left bundle branch block Structure of report This report is a systematic review update, based on a systematic review by Mowatt and colleagues (2008) [further detail provided in the Methods]. The Methods section describes the review s methods and includes the research questions, search strategy, inclusion and exclusion criteria, the data extraction, appraisal and synthesis methods, and the methodological limitations of the evidence review. The Results section considers the characteristics and results of the included appraised studies, and includes a meta-analysis of relevant outcomes where appropriate. Study characteristics and findings are reported in separate tables and synthesised in the text. The final section summarises results, briefly discusses the limitations of the evidence

16 4 base and identified gaps in knowledge, and presents key conclusions. Detailed appendices follow, including the search strategy, all excluded papers annotated by reason for exclusion, and the completed data extraction tables for included papers.

17 5 Methods The systematic review update was based on a systematic review provided by the requestor (Table 1). Table 1: Recent systematic review, selected for updating HTA organisation/authors Country Title Health Technology Assessment NHS R&D HTA Programme (Mowatt et al 2008) United Kingdom Systematic review of the clinical effectiveness and cost-effectiveness of 64-slice or higher computed tomography as an alternative to invasive coronary in the investigation of coronary artery disease Literature search end date December 2006 Abbreviations: HTA = health technology assessment; NHS = National Health Service; R&D = research and development The reviewer also identified an unpublished draft report released by the European Network for Health Technology Assessment (EUnetHTA) on CTA (2008). However the report was prepared to test the diagnostic HTA Core Model, rather than prepare a valid assessment on CTA. Thus the aforementioned systematic review by Mowatt and colleagues (2008) was deemed the most appropriate to be updated. The Mowatt et al (2008) systematic review concluded that the main value of 64-slice CT at present is to rule out significant CAD. It suggested it is unlikely to replace ICA in assessment for revascularisation of patients, particularly as and angioplasty are often done on the same occasion. The review also concluded that further research is needed into the advantages and costs of 256-slice machines compared with 64-slice CT. Research questions The clinical question to be answered by this review was defined by staff from The Sector Capability and Innovation Directorate of New Zealand s Ministry of Health in conjunction with the reviewers. In general, the aim of this review was to evaluate the recent evidence pertaining to the relative clinical effectiveness of 64-slice or higher CT as an alternative to ICA in the investigation of patients with suspected CAD. In accordance with standard procedure, the review question was defined according to the PICO (or PICOT) criteria: Patient population Intervention Comparator (where appropriate) Outcomes Time consideration (should be considered with regard to all of the above domains)

18 6 For inclusion in the current review, the evidence had to fulfil the criteria outlined in Table 2 and Table 3. These criteria were developed a priori and described in the scoping protocol prepared prior to commencement of the review proper. As opposed to the broader NHS systematic review, this review update did not include prognostic studies, technical studies (e.g. image quality), assessment studies or postrevascularisation studies. Table 2: Criteria for determining study eligibility Patient population Intervention Comparator Adult participants undergoing CT using 64-slice or higher machines for the detection of CAD. Several patient subgroups were also examined. Any studies including patients who had undergone a revascularisation procedure were excluded. Any studies examining asymptomatic patients were excluded. Patients must have had 64-slice or higher CT (single or dual source) and an ICA. Similar to the NHS review, MPS is a potential comparator, however studies that report results for 64-slice or higher CT without a non-invasive comparator will also be included. Outcomes Diagnostic accuracy studies had to report the absolute number of true positives, false positives, false negatives and true negatives, or sensitivity and specificity for CT in determining significant ( 50% stenosis) CAD. Abbreviations: CAD = coronary artery disease; CT = computed tomography; ICA = invasive coronary ; MPS = myocardial perfusion scintigraphy; NHS = National Health Service Table 3: Nature of the evidence Publication type Study design Study duration Studies published in the English language, including primary research published as full original reports. Conference abstracts, reviews, case reports, animal studies, short notes, letters and editorials were excluded. Randomised controlled trials or prospective/retrospective non-randomised comparative studies or case series in which some or all patients received both CTA using 64-slice or higher machines and ICA. Not specified. Only papers published from December 2006 onwards were included. Sample size Only studies where more than 50 participants underwent both 64-slice or higher CTA and ICA. Abbreviations: CTA = computed tomography ; ICA = invasive coronary Level of analysis Part of the criteria for determining studies eligibility was that articles reported either the absolute number of true positives, false positives, false negatives and true negatives, or sensitivity and specificity. Due to the nature of CTA, this could have been presented at the patient, vessel or segment level. Some papers reported diagnostic performance results for individual coronary arteries including the right coronary artery, left coronary artery, left anterior descending and circumflex artery. Most reported the results for all segments with some papers breaking this down further to proximal and distal segments, or side branches. The level of analysis has implications for clinical practice. Patient-level results are the most important from a patient management perspective, because if a significant stenosis is successfully detected, then usually that patient will be referred for an ICA. If results suggest no significant stenosis, then theoretically the patient would not require an ICA and would continue with conservative management. It should be noted that alternative approaches are available in clinical practice. For example, if CTA shows significant but relatively mild disease, such as a single stenosis in a mid or distal vessel, then an alternative approach is to perform further non-invasive testing to

19 7 look for evidence of ischaemia. If there is no ischaemia, then it is reasonable to defer ICA as it is unlikely that coronary revascularisation is required. At the vessel-level, if a significant stenosis is detected in a certain artery (i.e. left anterior descending artery), this will better inform the clinician who is to perform the forthcoming intervention. The segment-level analysis is valuable to gage the diagnostic precision of the technology and can also provide useful information for a re-vascularisation procedure if it is required. Literature search A systematic method of literature searching and selection was employed in the preparation of this review update. The NHS technology assessment report (Mowatt et al 2008) upon which this update was based had a search date end of December For this update, searches were limited to English language material published from December 2006 onwards. The searches were completed on the 25 th of March, Therefore, studies published after this date were not eligible for inclusion in the systematic review update. The following databases were searched: Embase Medline The Cochrane Library The reference lists of key papers were searched to identify any peer-reviewed evidence that may have been missed in the literature search. In accordance with the Scoping Protocol, hand-searching was not undertaken and the grey literature and unpublished material such as conference abstracts were not searched. Search terms were searched for as keywords, exploded where possible, and as free text within the title and/or abstract, in the Embase and Medline databases. Variations on these terms were used for the Cochrane Library, modified to suit their keywords and descriptors. The search terms, search strategy and number of citations identified are presented in Table 4.

20 8 Table 4: Search strategy Database EMBASE + MEDLINE (searched via Embase.com) Date(s) searched 01 Dec March 2009 Search no. Search terms #1 ((('myocardial ischemia'/exp OR 'myocardial ischemia') OR ('ischemic heart disease'/exp OR 'ischemic heart disease') OR ('coronary artery disease'/exp OR 'coronary artery disease') OR ('myocardial infarction'/exp OR 'myocardial infarction') OR ('chest pain'/exp OR 'chest pain') OR ('coronary artery blood flow'/exp OR 'coronary artery blood flow') OR ('angina'/exp OR 'angina') OR ('chest *3 pain')) OR ('ischemia *3 heart' OR ('ischemia *3 coronary') OR ('ischemia *3 myocardial' ) OR ('myocardial *3 thrombosis' ) OR ('myocardial *3 stenosis') OR ('myocardial *3 restenosis') OR ('myocardial *3 arteriosclerosis' ) OR ('myocardial *3 artery') OR ('myocardial *3 arteries') OR ('coronary *3 infarct') OR ('coronary *3 infarction') OR ('coronary *3 thrombosis') OR ('coronary *3 stenosis') OR ('coronary *3 restenosis') OR ('coronary *3 arteriosclerosis') OR ('coronary *3 artery') OR ('coronary *3 arteries') OR ('heart *3 arteries') OR ('heart *3 artery') OR ('heart *3 arteriosclerosis') OR ('heart *3 restenosis') OR ('heart *3 stenosis') OR ('heart *3 thrombosis') OR ('heart *3 infarct') OR ('heart *3 infarction') AND [english]/lim AND [humans]/lim AND [ ]/sd AND [ ]/py)) #2 ('tomography, x-ray computed') OR ('tomography, spiral computed') OR ('computer assisted tomography') OR ('computed tomographic ') OR ('electron beam tomography') OR ('high resolution computer tomography') OR ('spiral computer assisted tomography') OR ('detector' OR 'slice' OR 'slide' OR 'row') OR ('multidetector computed tomography') OR ('msct'/exp OR 'msct') OR ('mdct'/exp OR 'mdct') OR ('multi slice' OR 'multislice') OR ('multi row' OR 'multirow') OR ('multi slide' OR 'multislide') OR ('multi detect' OR 'multidetect') OR (('ct '/exp OR 'ct ') AND [english]/lim AND [humans]/lim AND [ ]/py) #3 (('coronary '/exp OR 'coronary ') OR ('coronary *3 angiograph$') OR ('coronary *3 angiogram$') OR ('coronary *3 arteriograph$') OR ('myocardial *3 angiogram') AND [english]/lim AND [humans]/lim AND [ ]/sd AND [ ]/py) #4 #1 AND #2 AND #3 1,483 Citations 83, ,874 7,457

21 9 Table 4: Search strategy (continued) Database Date(s) searched Search no. Search terms Citation s Cochrane Library #1 myocardial ischemia OR coronary disease OR myocardial infarction OR chest pain OR angina OR (ischemi* or ischaemi*) NEAR/3 (heart or coronary or myocardial) OR (chest NEAR/3 pain) OR (myocardial OR coronary OR heart) NEAR/3 (infarct* or thrombosis OR stenosis) OR (myocardial or heart or coronary) NEAR/3 (restenosis or arteriosclerosis or arter*):ti,ab,kw Clinical trials registry (Clinicaltrials.gov) 01/12/ /03/2009 #2 Tomography, X-Ray Computed, this term only OR Tomography, Spiral Computed, this term only AND (detector* or slice* or slide* or row*):ti,ab,kw #3 (msct or mdct):ti,ab,kw or (multislice* or multi slice*):ti,ab,kw or (multirow* or multi row*):ti,ab,kw or (multi detector* or multidetector*):ti,ab,kw or (multislide* or multi slide*):ti,ab,kw #4 (tomograph* near/5 (multi* or slice* or slide* or row* or detector*)):ti,ab,kw 186 #5 #2 OR #3 OR #4 1,756 #6 #1 AND #5, from 2006 to #7 Citations from clinical trials 36 #1 ct AND closed studies, intervention studies, received on or after 01/12/ Bibliographies of included studies and other sources 0 Consolidated number of citations from all sources 1,540 Non duplicate citations a 1, 438 a Duplicate citations removed manually using Reference Manager 29,497 1,

22 10 Assessment of study eligibility Studies were selected for appraisal using a two-stage process. First, titles and abstracts (where available) identified from the search strategy were scanned and excluded as appropriate. Second, the full text articles were retrieved for the remaining studies and selected for inclusion and appraisal in the review if they fulfilled the study selection criteria outlined below. Double-checking of the eligibility of studies by a second reviewer was not undertaken. As mentioned earlier, non-english publications were excluded at the database searching stage. Citations were excluded for the following reasons: 1. Not a diagnostic performance study: including prognostic studies, technical performance studies, case reports, animal studies, short notes, letters, editorials, conference abstracts, in-vitro studies, study types not deemed appropriate to the research question or nature of review 2. Wrong intervention: does not include the correct intervention/s (i.e. 64-slice or higher CT ) 3. Wrong patient group: does not include the correct patient group (studies with patients who had previously undergone a revascularisation procedure and population screening studies were excluded) 4. Wrong outcomes: does not include results (e.g. sensitivity, specificity, positive and negative predictive value) relating to the identified outcome of interest ( 50% stenosis) 5. Fewer than 50 study participants receiving both CTA and the reference standard There were 1,438 non-duplicate studies identified by the search strategy. As detailed in Table 5, 99 full text articles were eligible for retrieval after excluding studies from the search titles and abstracts. Of the full papers retrieved, 71 did not fulfil the inclusion criteria and were excluded. Therefore, 28 articles were fully appraised and are included in this report (listed in Appendix A). All excluded articles are presented in Appendix B, annotated by reason for exclusion based on the exclusion criteria detailed above. Reasons are presented hierarchically such that the first reason in the list that applied is reported. Other cited publications (e.g. those providing background material) are presented in the References.

23 11 Table 5: Application of selection criteria to citations Exclusion criteria Number Total citations 1438 Title/abstract: Not a diagnostic performance study: including prognostic studies, technical performance studies, case reports, animal studies, short notes, letters, editorials, conference abstracts, in-vitro studies, study types not deemed appropriate to the research question or nature of review Wrong intervention: does not include the correct intervention/s (i.e. 64-slice or higher CT ) Wrong patient group: does not include the correct patient group (studies with patients who had previously undergone a revascularisation procedure were excluded) Wrong outcomes: does not include results (e.g. sensitivity, specificity, positive and negative predictive value) relating to the identified outcome of interest ( 50% stenosis) Fewer than 50 study participants receiving both CTA and the reference standard Citations excluded after title/abstract review 1339 Full papers reviewed: 99 Full text: Not a diagnostic performance study: including prognostic studies, technical performance studies, case reports, animal studies, short notes, letters, editorials, conference abstracts, in-vitro studies, study types not deemed appropriate to the research question or nature of review Wrong intervention: does not include the correct intervention/s (i.e. 64-slice or higher CT ) Wrong patient group: does not include the correct patient group (studies with patients who had previously undergone a revascularisation procedure were excluded) Wrong outcomes: does not include results (e.g. sensitivity, specificity, positive and negative predictive value) relating to the identified outcome of interest ( 50% stenosis) Fewer than 50 study participants receiving both CTA and the reference standard (ICA) 5 Citations excluded after full text review 1410 Total included citations 28 Appraisal of included studies Dimensions of evidence The aim of this review was to find the highest quality evidence to answer the clinical question. In accordance with NHMRC guidance, the following dimensions of evidence were reviewed for each of the included studies (Table 6). It is important to recognise that the value of a piece of evidence is determined by all of these dimensions, not just the level of evidence

24 12 Table 6: Dimensions of evidence (NHMRC, 2000) Dimension Strength of evidence Level Quality Statistical precision Size of effect Relevance of evidence Definition The study design used, as an indicator of the degree to which bias has been eliminated by design. Each study will be assigned a level of evidence in accordance with the NHMRC (2005) diagnostic levels of evidence (Table 5). The methods used by the investigators to minimise bias within a study design. The P-value or alternatively, the precision of the estimate of the effect (as indicated by the confidence interval). It reflects the degree of certainty about the existence of a true effect. The distance of the study estimate from the null value and the inclusion of only clinically important effects in the confidence interval. The usefulness of the evidence in clinical practice, particularly the appropriateness of the outcome measures used. Each study was assigned a level of evidence in accordance with the NHMRC (2005) diagnostic levels of evidence (Table 7). The levels of evidence vary according to the nature of the research question. Importantly, the level of evidence is assigned at the individual study level, rather than to the body of evidence. Table 7: NHMRC diagnostic levels of evidence Level Diagnosis ** I * II III-1 III-2 A systematic review of level II studies A study of test accuracy with: an independent, blinded comparison with a valid reference standard, among consecutive patients with a defined clinical presentation A study of test accuracy with: an independent, blinded comparison with a valid reference standard, among non-consecutive patients with a defined clinical presentation A comparison with reference standard that does not meet the criteria required for Level II and III-1 evidence III-3 Diagnostic case-control study IV Study of diagnostic yield (no reference standard) * A systematic review will only be assigned a level of evidence as high as the studies it contains, excepting where those studies are of level II evidence. ** The dimensions of evidence apply only to studies of diagnostic accuracy. To assess the effectiveness of a diagnostic test there also needs to be a consideration of the impact of the test on patient management and health outcomes. See MSAC (2004) Guidelines for the assessment of diagnostic technologies. Available at: The validity of the reference standard should be determined in the context of the disease under review. Criteria for determining the validity of the reference standard should be pre-specified. This can include the choice of the reference standard(s) and its timing in relation to the index test. The validity of the reference standard can be determined through quality appraisal of the study. See Whiting P, Rutjes AWS, Reitsma JB, Bossuyt PMM, Kleijnen J. The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews. BMC Medical Research Methodology 2003, 3: 25. Well-designed population based case-control studies (e.g. population based screening studies where test accuracy is assessed on all cases, with a random sample of controls) do capture a population with a representative spectrum of disease and thus fulfil the requirements for a valid assembly of patients. However, in some cases the population assembled is not representative of the use of the test in practice. In diagnostic case-control studies a selected sample of patients already known to have the disease are compared with a separate group of normal/healthy people known to be free of the disease. In this situation patients with borderline or mild expressions of the disease, and conditions mimicking the disease are excluded, which can lead to exaggeration of both sensitivity and specificity. This is called spectrum bias because the spectrum of study participants will not be representative of patients seen in practice. Studies of diagnostic yield provide the yield of diagnosed patients, as determined by an index test, without confirmation of the accuracy of this diagnosis by a reference standard. These may be the only alternative when there is no reliable reference standard. Hierarchies adapted and modified from: NHMRC 1999; Bandolier 1999; Lijmer et al. 1999; Phillips et al In addition, in accordance with the NHS systematic review, individual study quality for this update was assessed using a modified version of the Quality Assessment of Diagnostic Studies (QUADAS) tool. Quality criteria is tabulated in the data extraction

25 13 form, rather than used to formulate a numeric score. The modified version of the QUADAS is shown below in Table 8. Table 8: Modified QUADAS quality assessment checklist (derived from QUADAS tool) Item Yes No Unclear Mandatory quality items 1. Was the spectrum of patients representative of the patients who will receive the test in practice? 2. Is the reference standard likely to correctly classify the target condition? Is the time period between the reference standard and index test short 3. enough to be reasonably sure that the target condition did not change between the two tests? a 4. Did the whole sample or a random selection of the sample receive verification using a reference standard of diagnosis? 5. Did patients receive the same reference standard regardless of the index test result? 6. Was the reference standard independent of the index test (i.e. the index test did not form part of the reference standard)? 7. Were the index test results interpreted without knowledge of the results of the reference standard? 8. Were the reference standard results interpreted without knowledge of the results of the index test? 9. Were the same clinical data available when test results were interpreted as would be available when the test is used in practice? b 10. Were uninterpretable/intermediate test results reported? 11. Were withdrawals from the study explained? Source: Mowatt et al (2008), adapted from Whiting et al (2003). a For Q3 to be checked Yes, the period should be less than 6 months. b For Q9 to be checked Yes, the data should be for CT results only. Data extraction Data were extracted onto specifically designed data extraction forms, and included information regarding study design, patient characteristics, details of the index test and reference standard, study quality and relevant results. Data were extracted by one reviewer. Completed data extraction forms regarding study characteristics, study quality (QUADAS) and observed diagnostic results can be found in Appendix C, Appendix D and Appendix E, respectively. A summary of the study characteristics are also presented in the results section of this report with a detailed description of calculated diagnostic performance (i.e. sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and overall diagnostic accuracy). Where available, results are reported for all levels of analysis (i.e. patient-level analysis, vessel-level analysis and segment-level analysis). Data synthesis In addition to the level and quality of evidence of individual studies, the review considers the body of evidence in total. This involves consideration of the volume of evidence and its consistency. The calculated diagnostic performance results of CTA at the patient-level were metaanalysed using Review Manager Version 5.0 and Metadisc Version 1.4 with two separate analyses included. The first meta-analysis, the base-case meta-analysis,

26 14 presents the diagnostic performance results of CTA at the patient-level but omits studies that excluded equivocal test results. The second meta-analysis, the alternative meta-analysis, included all studies that presented patient-level results with treatment of equivocal test results preferentially included using the following criteria: The first preference was for results (either as reported by authors or re-calculated by the reviewer) with equivocal CTA test results included and treated as test positive, with disease status as determined by ICA correctly assigned. The second preference was for results presented with equivocal test results treated as false positives (i.e. intent-to-diagnose principal). In this case the ICA result for equivocal segments was unknown and therefore could not be re-calculated. This convention assumes that the test was classified positive and the disease state negative. The third preference was for results presented with equivocal tests excluded. For patient-level analysis, forest plots were also included. Pooled sensitivity, specificity, PPV, NPV and diagnostic accuracy results were also calculated for: all vessels; the right coronary; left coronary; left anterior descending; and circumflex arteries, as well as all segments, using the equivocal test result selection methodology outlined above. Finally, the review considers the relevance of the evidence, both with regard to the applicability of the patient population and the intervention, as well as the relevance to the New Zealand health care setting. Limitations of the review methodology This review used a structured approach to review the literature. However, there were some inherent limitations with this approach. All types of study are subject to bias, with systematic reviews being subject to the same biases seen in the original studies they include, as well as biases specifically related to the systematic review process. Reporting biases are a particular problem related to systematic reviews and include publication bias, time-lag bias, multiple publication bias, language bias and outcome reporting bias (Egger et al. 2001). A brief summary of the different types of reporting bias is shown in Table 9. Other biases can result if the methodology to be used in a review is not defined a priori (i.e., before the review commences). Detailed knowledge of studies performed in the area of interest may influence the eligibility criteria for inclusion of studies in the review and may therefore result in biased results. For example, studies with more positive results may be preferentially included in a review, thus biasing the results and overestimating treatment effect.

27 15 Table 9: Reporting biases in systematic reviews * Type of bias Publication bias Time-lag bias Multiple publication bias Citation bias Language bias Outcome reporting bias Definition and effect on results of review The publication or non-publication of research findings. Small, negative trials tend not to be published and this may lead to an overestimate of results of a review if only published studies are included. The rapid or delayed publication of research findings. Studies with positive results tend to be published sooner than studies with negative findings and hence results may be overestimated until the negative trials catch up. The multiple or singular publication of research findings. Studies with significant results tend to be published multiple times which increases the chance of duplication of the same data and may bias the results of a review. The citation or non-citation of research. Citing of trials in publications is not objective so retrieving studies using this method alone may result in biased results. Unsupported studies tend to be cited often which may also bias results. The publication of research findings in a particular language. Significant results are more likely to be published in English so a search limited to English-language journals may result in an overestimation of effect. The selective reporting of some outcomes but not others. Outcomes with favourable findings may be reported more. For example, adverse events have been found to be reported more often in unpublished studies. This may result in more favourable results for published studies. * Adapted from Egger et al. (2001). Some of these biases are potentially present in this review. Only data published in peer-reviewed journals are included. No attempt was made to include unpublished material, as such material typically has insufficient information upon which to base quality assessment, and it has not been subject to the scrutiny of the peer-review process. In addition, the search was limited to English-language publications only so language bias is a potential problem also. Outcome reporting bias and inclusion criteria bias are unlikely as the reviewers had no detailed knowledge of the topic literature, and the methodology used in the review and the scope of the review was defined a priori. The review scope was developed with the assistance of Ministry of Health staff to support policy and purchasing relevant to New Zealand. All studies included in this review were conducted outside New Zealand, and therefore, their generalisability to the New Zealand population and context may be limited and needs to be considered. This review was confined to an examination of the efficacy of the intervention and did not consider ethical or legal considerations associated with this intervention. The studies were initially selected by examining the abstracts of these articles. Therefore, it is possible that some studies were inappropriately excluded prior to examination of the full text article. However, where detail was lacking, ambiguous papers were retrieved as full text to minimise this possibility. Reasons for exclusion for every article included in the review are presented in Appendix B for transparency. It should also be noted that where data were not reported in a two by two table, diagnostic performance results were based on information provided in text. The reader should be aware that this can result in errors due to misinterpretation of the text.

28 16 For a detailed description of interventions and evaluation methods, and results used in the studies appraised, the reader is referred to the original papers cited.

29 17 Results Overview Methodological information and results extracted from included studies are presented below. More detailed information is available in Appendix C, Appendix D and Appendix E, or in the original papers. Only data relevant to the current review is presented. All 28 papers identified as eligible for inclusion in the review were original research studies. The papers have been grouped together based on their assigned diagnostic level of evidence (i.e. Level II, Level III-1 and Level III-2). Included study characteristics The search strategy identified 28 relevant studies; nine assessed as Level II diagnostic evidence; 15 that were categorised as Level III-1; and four as Level III-2 diagnostic performance studies of CTA. The characteristics of the included studies are summarised in Table 10. Further details regarding included study characteristics are provided in Appendix C. The majority of studies were diagnostic intervention studies in which patients presenting with suspected CAD were prospectively analysed with 64-slice CTA to determine whether or not they had significant stenosis of coronary arteries (i.e. 50%). In order to validate the CTA results, all the included studies required patients to undergo a conventional ICA, which in some cases was performed prior to the 64- slice CTA. However in all studies the ICA was performed independently of the CTA, and with the exception of two studies, it was clear that the results of the index test (CTA) had been interpreted by a reviewer blinded to the results of the reference standard (ICA). Of the included studies, nine recruited patients consecutively whereas 17 either recruited patients non-consecutively or did not report the method of recruitment and were subsequently assumed to have recruited non-consecutively. Two included studies were retrospective analyses. In accordance with inclusion criteria determined a priori, each included study examined more than 50 patients with the number of participants analysed ranging from 51 to 402. Consistent with the gender and demographic profile for CAD, more males than females were examined in the included studies with mean age ranging from 53.9 to 68.2 years. The inclusion and exclusion criteria of the included studies were similar throughout. Although most studies did not explicitly report inclusion criteria, patients with suspected CAD because of a range of symptoms (e.g. angina), usually scheduled for ICA, were included in the studies. Those who had undergone a previous PCI such as stenting or CABG were excluded. Patients who had a contraindication to CTA such as a known allergy to iodinated contrast agent were also excluded. Other common exclusion criteria included atrial fibrillation, impaired renal function and inability to follow the breath hold command required to complete the test. The majority of patients were given beta-blockers prior to scanning in order to reduce their heart rate.