Stefan K. Lhachimi

A systematic review and economic evaluation of new-generation computed tomography scanners for imaging in coronary artery disease and congenital heart disease: Somatom Definition Flash, Aquilion ONE, Brilliance iCT and Discovery CT750 HD

BACKGROUND Computed tomography (CT) is important in diagnosing and managing many conditions, including coronary artery disease (CAD) and congenital heart disease. Current CT scanners can very accurately diagnose CAD requiring revascularisation in most patients. However, imaging technologies have developed rapidly and new-generation computed tomography (NGCCT) scanners may benefit patients who are difficult to image (e.g. obese patients, patients with high or irregular heart beats and patients who have high levels of coronary calcium or a previous stent or bypass graft). OBJECTIVE To assess the clinical effectiveness and cost-effectiveness of NGCCT for diagnosing clinically significant CAD in patients who are difficult to image using 64-slice computed tomography and treatment planning in complex congenital heart disease. DATA SOURCES Bibliographic databases were searched from 2000 to February/March 2011, including MEDLINE, MEDLINE In-Process and Other Non-Indexed Citations, EMBASE, Cochrane Database of Systematic Reviews (CDSR), Cochrane Central Register of Controlled Trials (CENTRAL), Database of Abstracts of Reviews of Effects (DARE), NHS Economic Evaluation Database (NHS EED), Health Technology Assessment (HTA) database and Science Citation Index (SCI). Trial registers and conference proceedings were searched. REVIEW METHODS Systematic review methods followed published guidance. Risk of bias was assessed using QUADAS-2. Results were stratified by patient group. Summary sensitivity and specificity were calculated using a bivariate summary receiver operating characteristic, or random effects model. Heterogeneity was assessed using the chi-squared statistic and I(2)-statistic. Cost-effectiveness of NGCCT was modelled separately for suspected and known CAD, evaluating invasive coronary angiography (ICA) only, ICA after positive NGCCT (NGCCT-ICA), and NGCCT only. The cost-effectiveness of NGCCT, compared with 64-slice CT, in reducing imaging-associated radiation in congenital heart disease was assessed. RESULTS Twenty-four studies reported accuracy of NGCCT for diagnosing CAD in difficult-to-image patients. No clinical effectiveness studies of NGCCT in congenital heart disease were identified. The pooled per-patient estimates of sensitivity were 97.7% [95% confidence interval (CI) 88.0% to 99.9%], 97.7% (95% CI 93.2% to 99.3%) and 96.0% (95% CI 88.8% to 99.2%) for patients with arrhythmias, high heart rates and previous stent, respectively. The corresponding estimates of specificity were 81.7% (95% CI 71.6% to 89.4%), 86.3% (95% CI 80.2% to 90.7%) and 81.6% (95% CI 74.7% to 87.3%), respectively. In patients with high coronary calcium scores, previous bypass grafts or obesity, only per-segment or per-artery data were available. Sensitivity estimates remained high (> 90% in all but one study). In patients with suspected CAD, the NGCCT-only strategy appeared most cost-effective; the incremental cost-effectiveness ratio (ICER) of NGCCT-ICA compared with NGCCT only was £71,000. In patients with known CAD, the most cost-effective strategy was NGCCT-ICA (highest cost saving, dominates ICA only). The ICER of NGCCT only compared with NGCCT-ICA was £726,230. For radiation exposure only, the ICER for NGCCT compared with 64-slice CT in congenital heart disease ranged from £521,000 for the youngest patients to £90,000 for adults. LIMITATIONS Available data were limited, particularly for obese patients and patients with previous bypass grafts. All studies of the accuracy of NGCCT assume that the reference standard (ICA) is 100% sensitive and specific; however, there is some evidence that ICA may sometimes underestimate the extent and severity of stenosis. Patients with more than one criterion that could contribute to difficulty in imaging were often excluded from studies; the effect on test accuracy of multiple difficult to image criteria remains uncertain. CONCLUSIONS NGCCT may be sufficiently accurate to diagnose clinically significant CAD in some or all difficult-to-image patient groups. Economic analyses suggest that NGCCT is likely to be considered cost-effective for difficult-to-image patients with CAD, at current levels of willingness to pay in the NHS. For patients with suspected CAD, NGCCT only would be most favourable; for patients with known CAD, NGCCT-ICA would be most favourable. No studies assessing the effects of NGCCT on therapeutic decision making, or subsequent patient outcomes, were identified. The ideal study to address these questions would be a large multi-centre RCT. However, one possible alternative might be to establish a multicentre tracker study. High-quality test accuracy studies, particularly in obese patients, patients with high coronary calcium, and those with previous bypass grafts are needed to confirm the findings of our systematic review. These studies should include patients with multiple difficult to image criteria. FUNDING The National Institute for Health Research Health Technology Assessment programme. This project was funded by the HTA programme, on behalf of NICE, as project number 10/107/01.

Health impacts of increasing alcohol prices in the European Union : a dynamic projection

OBJECTIVE Western Europe has high levels of alcohol consumption, with corresponding adverse health effects. Currently, a major revision of the EU excise tax regime is under discussion. We quantify the health impact of alcohol price increases across the EU. DATA AND METHOD We use alcohol consumption data for 11 member states, covering 80% of the EU-27 population, and corresponding country-specific disease data (incidence, prevalence, and case-fatality rate of alcohol related diseases) taken from the 2010 published Dynamic Modelling for Health Impact Assessment (DYNAMO-HIA) database to dynamically project the changes in population health that might arise from changes in alcohol price. RESULTS Increasing alcohol prices towards those of Finland (the highest in the EU) would postpone approximately 54,000 male and approximately 26,100 female deaths over 10 years. Moreover, the prevalence of a number of chronic diseases would be reduced: in men by approximately 97,800 individuals with diabetes, 65,800 with stroke and 62,200 with selected cancers, and in women by about 19,100, 23,500, and 27,100, respectively. CONCLUSION Curbing excessive drinking throughout the EU completely would lead to substantial gains in population health. Harmonisiation of prices to the Finnish level would, for selected diseases, achieve more than 40% of those gains.

Systematic Review of the Accuracy of Dual-Source Cardiac CT for Detection of Arterial Stenosis in Difficult to Image Patient Groups

PURPOSE To assess the diagnostic performance of dual-source cardiac (DSC) computed tomography (CT) newer-generation CT instruments for identifying anatomically significant coronary artery disease (CAD) in patients who are difficult to image by using 64-section CT. MATERIALS AND METHODS A literature search comprised bibliographic databases (January 1, 2000, to March 22, 2011, with a pragmatic update on September 6, 2012), trial registries, and conference proceedings. Only studies using invasive coronary angiography as reference standard were included. Risk of bias was assessed (QUADAS-2). Results were stratified according to patient group on the basis of clinical characteristics. Summary estimates of sensitivity and specificity of DSC CT for detecting 50% or greater arterial stenosis were calculated by using a bivariate summary receiver operating characteristic or random-effects model. RESULTS Twenty-five studies reported accuracy of DSC CT for diagnosing CAD in difficult to image patients; in 22 studies, one of two CT units of the same manufacturer (Somatom Definition or Somatom Definition Flash) was used, and in the remaining three, a different CT unit of another manufacturer (Aquilion One) was used. The pooled, per-patient estimates of sensitivity were 97.7% (95% confidence interval [CI]: 88.0%, 99.9%) and 97.7% (95% CI: 93.2%, 99.3%) for patients with arrhythmias and high heart rates, respectively. The corresponding pooled estimates of specificity were 81.7% (95% CI: 71.6%, 89.4%) and 86.3% (95% CI: 80.2%, 90.7%), respectively. All data were acquired by using Somatom Definition. In two studies with Somatom and one study with Aquilion One, sensitivity estimates of 90% or greater were reported in patients with previous stent implantations; specificities were 81.7% and 89.5% for Somatom and 81.0% for Aquilion One. In patients with high coronary calcium scores, previous bypass grafts, or obesity, only per-segment or per-artery data were available. Sensitivity estimates remained high (>90% in all but one study), and specificities ranged from 79.1% to 100%. All data were acquired by using Somatom Definition. CONCLUSION DSC CT may be sufficiently accurate to diagnose clinically significant CAD in some or all difficult to image patients. SUPPLEMENTAL MATERIAL http://radiology.rsna.org/lookup/suppl/doi:10.1148/radiol.13121136/-/DC1.

DYNAMO-HIA-A Dynamic Modeling Tool for Generic Health Impact Assessments

Background Currently, no standard tool is publicly available that allows researchers or policy-makers to quantify the impact of policies using epidemiological evidence within the causal framework of Health Impact Assessment (HIA). A standard tool should comply with three technical criteria (real-life population, dynamic projection, explicit risk-factor states) and three usability criteria (modest data requirements, rich model output, generally accessible) to be useful in the applied setting of HIA. With DYNAMO-HIA (Dynamic Modeling for Health Impact Assessment), we introduce such a generic software tool specifically designed to facilitate quantification in the assessment of the health impacts of policies. Methods and Results DYNAMO-HIA quantifies the impact of user-specified risk-factor changes on multiple diseases and in turn on overall population health, comparing one reference scenario with one or more intervention scenarios. The Markov-based modeling approach allows for explicit risk-factor states and simulation of a real-life population. A built-in parameter estimation module ensures that only standard population-level epidemiological evidence is required, i.e. data on incidence, prevalence, relative risks, and mortality. DYNAMO-HIA provides a rich output of summary measures – e.g. life expectancy and disease-free life expectancy – and detailed data – e.g. prevalences and mortality/survival rates – by age, sex, and risk-factor status over time. DYNAMO-HIA is controlled via a graphical user interface and is publicly available from the internet, ensuring general accessibility. We illustrate the use of DYNAMO-HIA with two example applications: a policy causing an overall increase in alcohol consumption and quantifying the disease-burden of smoking. Conclusion By combining modest data needs with general accessibility and user friendliness within the causal framework of HIA, DYNAMO-HIA is a potential standard tool for health impact assessment based on epidemiologic evidence.

The DYNAMO-HIA Model: An Efficient Implementation of a Risk Factor/Chronic Disease Markov Model for Use in Health Impact Assessment (HIA)

In Health Impact Assessment (HIA), or priority-setting for health policy, effects of risk factors (exposures) on health need to be modeled, such as with a Markov model, in which exposure influences mortality and disease incidence rates. Because many risk factors are related to a variety of chronic diseases, these Markov models potentially contain a large number of states (risk factor and disease combinations), providing a challenge both technically (keeping down execution time and memory use) and practically (estimating the model parameters and retaining transparency). To meet this challenge, we propose an approach that combines micro-simulation of the exposure information with macro-simulation of the diseases and survival. This approach allows users to simulate exposure in detail while avoiding the need for large simulated populations because of the relative rareness of chronic disease events. Further efficiency is gained by splitting the disease state space into smaller spaces, each of which contains a cluster of diseases that is independent of the other clusters. The challenge of feasible input data requirements is met by including parameter calculation routines, which use marginal population data to estimate the transitions between states. As an illustration, we present the recently developed model DYNAMO-HIA (DYNAMIC MODEL for Health Impact Assessment) that implements this approach.

Genetic, physiological, and lifestyle predictors of mortality in the general population

OBJECTIVES We investigated the quality of 162 variables, focusing on the contribution of genetic markers, used solely or in combination with other characteristics, when predicting mortality. METHODS In 5974 participants from the Rotterdam Study, followed for a median of 15.1 years, 7 groups of factors including age and gender, genetics, socioeconomics, lifestyle, physiological characteristics, prevalent diseases, and indicators of general health were related to all-cause mortality. Genetic variables were identified from 8 genome-wide association scans (n = 19,033) and literature review. RESULTS We observed 3174 deaths during follow-up. The fully adjusted model (C-statistic for 15-year follow-up [C15y] = 0.80; 95% confidence interval [CI] = 0.79, 0.81) predicted mortality well [corrected]. Most of the additional information apart from age and sex stemmed from physiological markers, prevalent diseases, and general health. Socioeconomic factors and lifestyle contributed meaningfully to mortality risk prediction with longer prediction horizon. Although specific genetic factors were independently associated with mortality, jointly they contributed little to mortality prediction (C(15y) = 0.56; 95% CI = 0.55, 0.57). CONCLUSIONS Mortality can be predicted reasonably well over a long period. Genetic factors independently predict mortality, but only modestly more than other risk indicators.

Standard tool for quantification in health impact assessment a review.

BACKGROUND The health impact assessment (HIA) of policy proposals is becoming common practice. HIA represents a broad approach with quantification of the impact of policy options at its core. However, no standard tool is available and it remains unclear whether any current model can serve as a standard for the field. PURPOSE The aim of this study is to assess whether already existing models can be used as a standard tool for the quantification step in an HIA. METHODS A search in 2008 identified 20 models for HIA, of which six are sufficiently generic to allow for various and multiple diseases and different risk factors: Age-Related Morbidity and Death Analysis, Global Burden of Disease, Population Health Modeling, PREVENT, Proportional Life Table Method, and the National Institute for Public Health and the Environment (the Netherlands) Chronic Disease Model. These were evaluated along three proposed model structure criteria (real-life population, dynamic projection, explicit risk-factor states) and three usability criteria (modest data requirements, rich model output, generally accessible) developed to address the needs and requirements of the HIA framework. RESULTS Of the six generic models investigated, none fulfills all the proposed criteria as a standard HIA tool. The models are either technically advanced with no or limited accessibility, or they are accessible but oversimplified. CONCLUSIONS Further work on models for HIA with equal emphasis on technical appropriateness, availability of data, and end-user-friendly implementation is warranted if the field is to move forward.

Comparison of Tobacco Control Scenarios: Quantifying Estimates of Long-Term Health Impact Using the DYNAMO-HIA Modeling Tool

Background There are several types of tobacco control interventions/policies which can change future smoking exposure. The most basic intervention types are 1) smoking cessation interventions 2) preventing smoking initiation and 3) implementation of a nationwide policy affecting quitters and starters simultaneously. The possibility for dynamic quantification of such different interventions is key for comparing the timing and size of their effects. Methods and Results We developed a software tool, DYNAMO-HIA, which allows for a quantitative comparison of the health impact of different policy scenarios. We illustrate the outcomes of the tool for the three typical types of tobacco control interventions if these were applied in the Netherlands. The tool was used to model the effects of different types of smoking interventions on future smoking prevalence and on health outcomes, comparing these three scenarios with the business-as-usual scenario. The necessary data input was obtained from the DYNAMO-HIA database which was assembled as part of this project. All smoking interventions will be effective in the long run. The population-wide strategy will be most effective in both the short and long term. The smoking cessation scenario will be second-most effective in the short run, though in the long run the smoking initiation scenario will become almost as effective. Interventions aimed at preventing the initiation of smoking need a long time horizon to become manifest in terms of health effects. The outcomes strongly depend on the groups targeted by the intervention. Conclusion We calculated how much more effective the population-wide strategy is, in both the short and long term, compared to quit smoking interventions and measures aimed at preventing the initiation of smoking. By allowing a great variety of user-specified choices, the DYNAMO-HIA tool is a powerful instrument by which the consequences of different tobacco control policies and interventions can be assessed.

Communicating the parameter uncertainty in the IQWiG efficiency frontier to decision-makers.

The Institute for Quality and Efficiency in Health Care (IQWiG) developed—in a consultation process with an international expert panel—the efficiency frontier (EF) approach to satisfy a range of legal requirements for economic evaluation in Germanys statutory health insurance system. The EF approach is distinctly different from other health economic approaches. Here, we evaluate established tools for assessing and communicating parameter uncertainty in terms of their applicability to the EF approach. Among these are tools that perform the following: (i) graphically display overall uncertainty within the IQWiG EF (scatter plots, confidence bands, and contour plots) and (ii) communicate the uncertainty around the reimbursable price. We found that, within the EF approach, most established plots were not always easy to interpret. Hence, we propose the use of price reimbursement acceptability curves—a modification of the well-known cost-effectiveness acceptability curves. Furthermore, it emerges that the net monetary benefit allows an intuitive interpretation of parameter uncertainty within the EF approach. This research closes a gap for handling uncertainty in the economic evaluation approach of the IQWiG methods when using the EF. However, the precise consequences of uncertainty when determining prices are yet to be defined.

Modelling obesity outcomes: reducing obesity risk in adulthood may have greater impact than reducing obesity prevalence in childhood

A common policy response to the rise in obesity prevalence is to undertake interventions in childhood, but it is an open question whether this is more effective than reducing the risk of becoming obese during adulthood. In this paper, we model the effect on health outcomes of (i) reducing the prevalence of obesity when entering adulthood; (ii) reducing the risk of becoming obese throughout adult life; and (iii) combinations of both approaches. We found that, while all approaches reduce the prevalence of chronic diseases and improve life expectancy, a given percentage reduction in obesity prevalence achieved during childhood had a smaller effect than the same percentage reduction in the risk of becoming obese applied throughout adulthood. A small increase in the probability of becoming obese during adulthood offsets a substantial reduction in prevalence of overweight/obesity achieved during childhood, with the gains from a 50% reduction in child obesity prevalence offset by a 10% increase in the probability of becoming obese in adulthood. We conclude that both policy approaches can improve the health profile throughout the life course of a cohort, but they are not equivalent, and a large reduction in child obesity prevalence may be reversed by a small increase in the risk of becoming overweight or obese in adulthood.