Systematic review and modelling of the cost-effectiveness of cardiac magnetic resonance imaging compared with current existing testing pathways in ischaemic cardiomyopathy

BACKGROUND: Cardiac magnetic resonance imaging (CMR) is increasingly used to assess patients for myocardial viability prior to revascularisation. This is important to ensure that only those likely to benefit are subjected to the risk of revascularisation. OBJECTIVES: To assess current evidence on the accuracy and cost-effectiveness of CMR to test patients prior to revascularisation in ischaemic cardiomyopathy; to develop an economic model to assess cost-effectiveness for different imaging strategies; and to identify areas for further primary research. DATA SOURCES: Databases searched were: MEDLINE including MEDLINE In-Process & Other Non-Indexed Citations Initial searches were conducted in March 2011 in the following databases with dates: MEDLINE including MEDLINE In-Process & Other Non-Indexed Citations via Ovid (1946 to March 2011); Bioscience Information Service (BIOSIS) Previews via Web of Science (1969 to March 2011); EMBASE via Ovid (1974 to March 2011); Cochrane Database of Systematic Reviews via The Cochrane Library (1996 to March 2011); Cochrane Central Register of Controlled Trials via The Cochrane Library 1998 to March 2011; Database of Abstracts of Reviews of Effects via The Cochrane Library (1994 to March 2011); NHS Economic Evaluation Database via The Cochrane Library (1968 to March 2011); Health Technology Assessment Database via The Cochrane Library (1989 to March 2011); and the Science Citation Index via Web of Science (1900 to March 2011). Additional searches were conducted from October to November 2011 in the following databases with dates: MEDLINE including MEDLINE In-Process & Other Non-Indexed Citations via Ovid (1946 to November 2011); BIOSIS Previews via Web of Science (1969 to October 2011); EMBASE via Ovid (1974 to November 2011); Cochrane Database of Systematic Reviews via The Cochrane Library (1996 to November 2011); Cochrane Central Register of Controlled Trials via The Cochrane Library (1998 to November 2011); Database of Abstracts of Reviews of Effects via The Cochrane Library (1994 to November 2011); NHS Economic Evaluation Database via The Cochrane Library (1968 to November 2011); Health Technology Assessment Database via The Cochrane Library (1989 to November 2011); and the Science Citation Index via Web of Science (1900 to October 2011). Electronic databases were searched March-November 2011. REVIEW METHODS: The systematic review selected studies that assessed the clinical effectiveness and cost-effectiveness of CMR to establish the role of CMR in viability assessment compared with other imaging techniques: stress echocardiography, single-photon emission computed tomography (SPECT) and positron emission tomography (PET). Studies had to have an appropriate reference standard and contain accuracy data or sufficient details so that accuracy data could be calculated. Data were extracted by two reviewers and discrepancies resolved by discussion. Quality of studies was assessed using the QUADAS II tool (University of Bristol, Bristol, UK). A rigorous diagnostic accuracy systematic review assessed clinical and cost-effectiveness of CMR in viability assessment. A health economic model estimated costs and quality-adjusted life-years (QALYs) accrued by diagnostic pathways for identifying patients with viable myocardium in ischaemic cardiomyopathy with a view to revascularisation. The pathways involved CMR, stress echocardiography, SPECT, PET alone or in combination. Strategies of no testing and revascularisation were included to determine the most cost-effective strategy. RESULTS: Twenty-four studies met the inclusion criteria. All were prospective. Participant numbers ranged from 8 to 52. The mean left ventricular ejection fraction in studies reporting this outcome was 24-62%. CMR approaches included stress CMR and late gadolinium-enhanced cardiovascular magnetic resonance imaging (CE CMR). Recovery following revascularisation was the reference standard. Twelve studies assessed diagnostic accuracy of stress CMR and 14 studies assessed CE CMR. A bivariate regression model was used to calculate the sensitivity and specificity of CMR. Summary sensitivity and specificity for stress CMR was 82.2% [95% confidence interval (CI) 73.2% to 88.7%] and 87.1% (95% CI 80.4% to 91.7%) and for CE CMR was 95.5% (95% CI 94.1% to 96.7%) and 53% (95% CI 40.4% to 65.2%) respectively. The sensitivity and specificity of PET, SPECT and stress echocardiography were calculated using data from 10 studies and systematic reviews. The sensitivity of PET was 94.7% (95% CI 90.3% to 97.2%), of SPECT was 85.1% (95% CI 78.1% to 90.2%) and of stress echocardiography was 77.6% (95% CI 70.7% to 83.3%). The specificity of PET was 68.8% (95% CI 50% to 82.9%), of SPECT was 62.1% (95% CI 52.7% to 70.7%) and of stress echocardiography was 69.6% (95% CI 62.4% to 75.9%). All currently used diagnostic strategies were cost-effective compared with no testing at current National Institute for Health and Care Excellence thresholds. If the annual mortality rates for non-viable patients were assumed to be higher for revascularised patients, then testing with CE CMR was most cost-effective at a threshold of £20,000/QALY. The proportion of model runs in which each strategy was most cost-effective, at a threshold of £20,000/QALY, was 40% for CE CMR, 42% for PET and 16.5% for revascularising everyone. The expected value of perfect information at £20,000/QALY was £620 per patient. If all patients (viable or not) gained benefit from revascularisation, then it was most cost-effective to revascularise all patients. LIMITATIONS: Definitions and techniques assessing viability were highly variable, making data extraction and comparisons difficult. Lack of evidence meant assumptions were made in the model leading to uncertainty; differing scenarios were generated around key assumptions. CONCLUSIONS: All the diagnostic pathways are a cost-effective use of NHS resources. Given the uncertainty in the mortality rates, the cost-effectiveness analysis was performed using a set of scenarios. The cost-effectiveness analyses suggest that CE CMR and revascularising everyone were the optimal strategies. Future research should look at implementation costs for this type of imaging service, provide guidance on consistent reporting of diagnostic testing data for viability assessment, and focus on the impact of revascularisation or best medical therapy in this group of high-risk patients. FUNDING: The National Institute of Health Technology Assessment programme.