Search

Given the advantages in transparency, reproducibility, adaptability and computational efficiency in R, there is a growing interest in converting existing spreadsheet-based models into an R script for model re-use and upskilling training among health economic modellers. The objective of this exercise was to convert the Scottish Cardiovascular Disease (CVD) Policy Model from Excel to R and discuss the lessons learnt throughout this process. The CVD model is a competing risk state transition cohort model. Four health economists, with varied experience of R, attempted to replicate an identical model structure in R based on the model in Excel and reproduce the intermediate and final results. Replications varied in their use of specialist health economics packages in addition to standard data management packages. Two versions of the CVD model were created in R along with a Shiny app. Version 1 was developed without health economics specialist packages and produced identical results to the Excel version. Version 2 used the heemod package and did not achieve the same results, possibly due to the non-standard elements of the model and limited time to adapt the functions. The R model requires less than half the computational time than the Excel model. Conversion of the spreadsheet models to script models is feasible for health economists. A step-by-step guide for the conversion process is provided and modellers' experience is discussed. Coding without specialist packages allows full flexibility, while specialist packages may add convenience if the model structure is suitable. Whichever approach is taken, transparency and replicability remain the key criteria in model programming. Model conversions must maintain standards in these areas regardless of the choice of software.

Background The CULPRIT-SHOCK trial compared two treatment strategies for patients with acute myocardial infarction and multivessel coronary artery disease complicated by cardiogenic shock: (a) culprit vessel only percutaneous coronary intervention (CO-PCI), with additional staged revascularisation if indicated, and (b) immediate multivessel PCI (MV-PCI). Methods A German societal and national health service perspective was considered for three different analyses. The cost utility analysis (CUA) estimated costs and quality adjusted life years (QALYs) based on a pre-trial decision analytic model taking a lifelong time horizon. In addition, a within trial CUA estimated QALYs and costs for 1 year. Finally, the cost effectiveness analysis (CEA) used the composite primary outcome, mortality and renal failure at 30-day follow-up, and the within trial costs. Econometric and survival analysis on the trial data was used for the estimation of the model parameters. Subgroup analysis was performed following an economic protocol. Results The lifelong CUA showed an incremental cost effectiveness ratio (ICER), CO-PCI vs. MV-PCI, of €7010 per QALY and a probability of CO-PCI being the most cost-effective strategy > 64% at a €30,000 threshold. The ICER for the within trial CUA was €14,600 and the incremental cost per case of death/renal failure avoided at 30-day follow-up was €9010. Cost-effectiveness improved with patient age and for those without diabetes. Conclusions The estimates of cost-effectiveness for CO-PCI vs. MV-PCI have been shown to change depending on the time horizon and type of economic evaluation performed. The results favoured a long-term horizon analysis for avoiding underestimation of QALY gains from the CO-PCI arm.

Given the advantages in transparency, reproducibility, adaptability and computational efficiency in R, there is a growing interest in converting existing spreadsheet-based models into an R script for model re-use and upskilling training among health economic modellers. The objective of this exercise was to convert the Scottish Cardiovascular Disease (CVD) Policy Model from Excel to R and discuss the lessons learnt throughout this process. The CVD model is a competing risk state transition cohort model. Four health economists, with varied experience of R, attempted to replicate an identical model structure in R based on the model in Excel and reproduce the intermediate and final results. Replications varied in their use of specialist health economics packages in addition to standard data management packages. Two versions of the CVD model were created in R along with a Shiny app. Version 1 was developed without health economics specialist packages and produced identical results to the Excel version. Version 2 used the heemod package and did not achieve the same results, possibly due to the non-standard elements of the model and limited time to adapt the functions. The R model requires less than half the computational time than the Excel model. Conversion of the spreadsheet models to script models is feasible for health economists. A step-by-step guide for the conversion process is provided and modellers' experience is discussed. Coding without specialist packages allows full flexibility, while specialist packages may add convenience if the model structure is suitable. Whichever approach is taken, transparency and replicability remain the key criteria in model programming. Model conversions must maintain standards in these areas regardless of the choice of software.

Earlier detection of lung cancer is possible, but difficult and costly to achieve. Screening with Low Dose Computed Tomography (LDCT)scanning has been shown to reduce mortality by 20-25% over the past decade but uptake amongst those most likely to suffer the disease has been slow. Resource constraints and a high false positive rate have also limited adoption of LDCT in many health systems. Targeted screening of people most likely to benefit using a range of biomarkers may be one way to improve the yield and reduce the resource requirements of LDCT. Autoantibodies, which amplify the signal produced by cancer derived proteins, are present in the blood of people mounting an immune response to cancer are a potential way to select those at highest risk. We have followed up 12 208 people enrolled in the ECLS trial for three years and shown that the specificity for early stage (I &II) disease is 90.3% throughout that period. More cancers were detected in the control than the intervention arm of the trial (101V 83). Sensitivity was 77.8% after 6 months and dropped to 46.4% after 3 years. At the end of three years the hazard ratios (95%CI) for All Cause, Cancer Specific and Lung Cancer Mortality was 0.82(0.67-1.01), 0.72(0.54-0.97) and 0.70(0.46-1.08) respectively for those randomised to Early CDT testing. As a range of treatment modalities become increasingly more effective it is even more important to target LDCT on those most likely to have early stage disease. Autoantibody testing may be one method of targeting early detection on those most likely to benefit.

This study widens the transferability of cost-utility results from the SCOT trial showing that administering 3 months of adjuvant, oxaliplatin-based chemotherapy is cost-effective and cost saving compared to 6 months from the perspective of all countries recruited to SCOT. The impact on healthcare budgets if the findings are implemented as predicted will amount to savings of at least US$150 million over 5 years. Background: The Short Course Oncology Treatment (SCOT) trial demonstrated non-inferiority, less toxicity, and cost-effectiveness from a UK perspective of 3 versus 6 months of oxaliplatin-based chemotherapy for patients with colorectal cancer. This study assessed the cost-effectiveness of shorter treatment and the budget impact of implementing trial findings from the perspectives of all countries recruited to SCOT: Australia, Denmark, New Zealand, Spain, Sweden, and the United Kingdom. Patients and Methods: Individual cost-utility analyses were performed from the perspective of each country. Resource, quality of life, and survival estimates from the SCOT trial (N = 6065) were used. Probabilistic sensitivity analysis and subgroup analyses were undertaken. Using undiscounted costs from these cost-utility analyses, the impact on country-specific healthcare budgets of implementing the SCOT trial findings was calculated over a 5-year period. The currency used was US dollars (US$), and 2019 was the base year. One-way and scenario sensitivity analysis addressed uncertainty within the budget impact analysis. Results: Three months of treatment were cost saving and cost-effective compared to 6 months from the perspective of all countries. The incremental net monetary benefit per patient ranged from US$8972 (Spain) to US$13,884 (Denmark). The healthcare budget impact over 5 years for the base-case scenario ranged from US$3.6 million (New Zealand) to US$61.4 million (UK) and totaled over US$150 million across all countries. Conclusion: This study has widened the transferability of results from the SCOT trial, showing that shorter treatment is cost-effective from a multi-country perspective. The vast savings from implementation could fully justify the investment in conducting the SCOT trial.

7076 Background: The international Short Course Oncology Treatment (SCOT) trial demonstrated non-inferiority and significantly less toxicity of 3 versus 6 months of adjuvant chemotherapy for patients with colorectal cancer (CRC). This study assesses the value of shorter treatment and the economic implications of implementing the findings from the perspective of the countries that participated in the SCOT trial. Methods: Individual patient level data (n=6055) from the SCOT trial was used in a fully-pooled, cost utility analysis for the six participating countries. The incremental net monetary benefit (INMB) per patient was calculated using a willingness to pay threshold of one Gross Domestic Product per capita for each country. Responses to a clinician questionnaire (n=265 across 21 countries collected in April 2019) were used to estimate extent of practice change. The budget impact over 5 years of using shorter treatment was calculated, using 2019 and US dollars (USD) as the base year and currency, respectively. Results: Cost drivers for differences between the SCOT trial arms were reduced chemotherapy costs and fewer hospitalisations in the first treatment year. The INMB per patient of using shorter treatment and subsequent monetary impact on healthcare provider budgets resulting from implementation are shown in Table. This is a cost saving treatment strategy in all countries. The budget impact over 5 years amounts to savings of nearly half a billion USD. Conclusions: The economic burden of CRC treatment globally exceeds $39 billion per annum. Understanding the costs and consequences of widespread clinical practice change is important for optimal budget planning. This study has widened the transferability of results from a phase III cancer trial, showing shorter treatment is cost-effective from a multi-country perspective. The vast savings could provide benefit elsewhere within a limited healthcare budget, and justify the investment in conducting the SCOT trial. [Table: see text]

e19341 Background: The Short Course Oncology Treatment (SCOT) trial demonstrated that three months of adjuvant, doublet chemotherapy for patients with colorectal cancer was non-inferior and significantly less toxic than six months. To understand if and how these results were implemented, this study analysed prescribing practice before and after dissemination of the SCOT results at a population level. Methods: Individual electronic records for patients in a Scottish health board (population 1.1 million) were reviewed (January 2010-June 2019) to explore the proportion of patients receiving over three months of treatment pre and post-SCOT. Logistic and linear regression were used in an interrupted time series analysis to estimate prescribing practices, and to assess the counterfactual situation in which the trial had not occurred. Patients receiving single agent chemotherapy were the comparator group. Predictors of patients receiving a longer duration of treatment post-SCOT were also identified. Results: In total, 998 patients received adjuvant chemotherapy; 564 doublet and 434 single agent. For those prescribed doublet chemotherapy, there was a significant decrease in the proportion receiving more than 3 months of treatment after the SCOT trial results (20%) versus pre-SCOT (84%, p < 0.001). Disease risk was a key predictor for longer duration of treatment post-SCOT. Table outlines the predicted practice post-SCOT (June 2019) for the modelled and counterfactual scenarios. Conclusions: A clear change in treatment duration for patients receiving doublet chemotherapy was observed. Disease stage influencing post-SCOT prescribing aligns with recently updated clinical guidelines. This methodology can be applied to explore the implementation of future trials and to highlight any barriers to implementation that exist. [Table: see text]

IntroductionEmergency percutaneous coronary intervention (PCI) of the culprit lesion for patients with acute myocardial infarctions is an accepted practice. A majority of patients present with multivessel disease with additional relevant stenoses apart from the culprit lesion. In haemodynamically stable patients, there is increasing evidence from randomised trials to support the practice of immediate complete revascularisation. However, in the presence of cardiogenic shock, the optimal management strategy for additional non-culprit lesions is unknown. A multicentre randomised controlled trial, CULPRIT-SHOCK, is examining whether culprit vessel only PCI with potentially subsequent staged revascularisation is more effective than immediate multivessel PCI. This paper describes the intended economic evaluation of the trial.Methods and analysisThe economic evaluation will be conducted using a pre-trial decision model and within-trial analysis. The modelling-based analysis will provide expected costs and health outcomes, and incremental cost-effectiveness ratio over the lifetime for the cohort of patients included in the trial. The within-trial analysis will provide estimates of cost per life saved at 30 days and in 1 year, and estimates of health-related quality of life. Bootstrapping and cost-effectiveness acceptability curves will be used to address any uncertainty around these estimates. Different types of regression models within a generalised estimating equation framework will be used to examine how the total cost and quality-adjusted life years are explained by patients’ characteristics, revascularisation strategy, country and centre. The cost-effectiveness analysis will be from the perspective of each country’s national health services, where costs will be expressed in euros adjusted for purchasing power parity.Ethics and disseminationEthical approval for the study was granted by the local Ethics Committee at each recruiting centre. The economic evaluation analyses will be published in peer-reviewed journals of the concerned literature and communicated through the profiles of the authors atwww.twitter.comandwww.researchgate.net.Trial registration numberNCT01927549; Pre-results.

We present data of a contingent valuation survey, testing the effect of evaluation mode on the monetary valuation of preventing road accidents. Half of the interviewees was asked to state their willingness to pay (WTP) to reduce the risk of having only 1 type of injury (separate evaluation, SE), and the other half of the sample was asked to state their WTP for 4 types of injuries evaluated simultaneously (joint evaluation, JE). In the SE group, we observed lack of sensitivity to scope while in the JE group WTP increased with the severity of the injury prevented. However, WTP values in this group were subject to context effects. Our results suggest that the traditional explanation of the disparity between SE and JE, namely, the so-called "evaluability," does not apply here. The paper presents new explanations based on the role of preference imprecision.

The alternation bias is the tendency of people to believe that random events alternate more often than statistical laws imply. This paper examines the theoretical effect of this psychological bias on preferences over repeated investments by using a model of the belief in the law of small numbers. An alternation bias agent (ABA) has a different perception to a rational agent (RA) about the outcome distribution of the sum of n realisations of a lottery. The results show that an ABA, that maximises expected utility, could reject a single realisation of a lottery while accepting several repetitions in accordance with Paul Samuelson's fallacy of large numbers. Furthermore, the explanation of this type of preference, based on the alternation bias, is compatible with previous behavioural accounts. A more general result shows that the alternation bias increases (decreases) the expected utility of the perceived sum of identically distributed lotteries if individuals are risk averse (risk seekers).

We analyze data of a Spanish nationally-representative survey where subjects reported their willingness to pay (WTP) for road safety improvements, specifically they hypothetically paid for a reduction of the risk of a road fatality and several injuries. Respondents also reported their current income (CI) and permanent income (PI). The latter refers to their normal income once they considered various stages of low/high earnings throughout their entire lives. Consequently, we define relative income as the comparison of CI with respect to PI. Three income frames are generated as explanatory variables: gain (with CIPI); neutral (with CI=PI); and loss scenario (with CIPI). Surprisingly, we find that conditional on current income, and on a set of characteristics, those respondents in gain frame reported higher WTP than those in neutral and loss scenario. Further analysis shows that the income frames effect is higher and more significant for the older half-sample (45), being about three or four times higher than for the younger subset. Possible interpretations of the role of PI as a reference point are considered given the results. A reference-dependent utility function of income, where PI is the reference point, is proposed to describe the monetary valuation of safety within the theoretical framework previously developed in the safety economics literature.