UK cost-effectiveness analysis of endoscopic sleeve gastroplasty versus lifestyle modification alone for adults with class II obesity

Background Endoscopic sleeve gastroplasty (ESG) is a minimally invasive procedure that has been demonstrated in the MERIT randomised, controlled trial to result in substantial and durable additional weight loss in adults with obesity compared with lifestyle modification (LM) alone. We sought to conduct the first cost-effectiveness analysis of ESG versus LM alone in adults with class II obesity (BMI 35.0–39.9 kg/m2) from a national healthcare system perspective in England based on results from this study. Methods A 6-state Markov model was developed comprising 5 BMI-based health states and an absorbing death state. Baseline characteristics, utilities, and transition probabilities were informed by patient-level data from the subset of patients with class II obesity in MERIT. Adverse events (AEs) were based on the MERIT safety population. Mortality was estimated by applying BMI-specific hazard ratios from the published literature to UK general population mortality rates. Utilities for the healthy weight and overweight health states were informed from the literature; disutility associated with increasing BMI in the class I-III obesity health states was estimated using MERIT utility data. Disutility due to AEs and the prevalence of obesity-related comorbidities were based on the literature. Costs included intervention costs, AE costs, and comorbidity costs. Results ESG resulted in higher overall costs than LM alone but led to an increase in quality-adjusted life years (QALYs). The incremental cost-effectiveness ratio (ICER) for ESG vs LM alone was £2453/QALY gained. ESG was consistently cost effective across a wide range of sensitivity analyses, with no ICER estimate exceeding £10,000/QALY gained. In probabilistic sensitivity analysis, the mean ICER was £2502/QALY gained and ESG remained cost effective in 98.25% of iterations at a willingness-to-pay threshold of £20,000/QALY. Conclusion Our study indicates that ESG is highly cost effective versus LM alone for the treatment of adults with class II obesity in England.


Supplementary Appendix
Table S1: Consolidated Health Economic Evaluation Reporting Standards (CHEERS) checklist ..... Table S2: Health state utility estimates based on MERIT study data and published literature .......... Table S3: Health state utility linear mixed-effects model results .a Based on the approach taken in NICE's appraisal of liraglutide for the treatment of obesity [11].
NHS National Health Service.

Table S1 : Consolidated Health Economic Evaluation Reporting Standards (CHEERS) checklist CHEERS domain Description Location in article
24 Effect of uncertainty Describe how uncertainty about analytic judgments, inputs, or projections affects findings.Report the effect of choice of discount rate and time horizon, if applicable.Report on any difference patient/service recipient, general public, community, or stakeholder involvement made to the approach or findings of the study.Methods 26 Study findings, limitations, generalizability, and current knowledge Report key findings, limitations, ethical, or equity considerations not captured and how these could impact patients, policy, or practice.Results; Discussion 27 Study funding Describe how the study was funded and any role of the funder in the identification, design, conduct, and reporting of the analysis.Note: Checklist published by Husereau et al, 2022 [1].a Model parameter sheet provided as a standalone Microsoft ® Excel file.

Table S2 : Health state utility estimates based on MERIT study data and published literature Based on MERIT study data a Model health state Number of SF-36 observations EQ-5D value (Rowen et al.) b EQ-5D value (Ara et al.) c Health state utility value reported by Stephenson et al.
bMapping algorithm used in the base-case analysis.cMapping algorithm explored in a model scenario analysis.EQ-5D EuroQol five dimensions health survey, SF-36 36-item short form health survey.

Table S3 : Health state utility linear mixed-effects model results
[3]e: Results from a linear mixed-effects regression model used to assess the incremental difference between the overweight health state utility value (0.81) reported by Stephenson et al, 2021[5]and the EQ-5D values for the obesity I, obesity II, and obesity III model health states mapped from MERIT SF-36 data using algorithms published by Rowen et al, 2009[3]andAra and Brazier, 2008 [4].aMapping algorithm used in the base-case analysis.bMapping algorithm explored in a model scenario analysis.EQ-5D EuroQol five dimensions health survey, SF-36 36-item short form health survey.

Table S4 : Health state utility estimates for the obesity I, II, and III model health states based on results from the linear mixed-effects model Utility estimates based on MERIT SF-36 data mapped to EQ-5D Model health state Algorithm from Rowen et al, 2009 b Algorithm from Ara and Brazier, 2008 c Utility reported by Stephenson et al, 2021
[5]36 data from MERIT (class II obesity subpopulation)[2], mapped to EQ-5D using algorithms published by Rowen et al, 2009[3]and Ara and Brazier, 2008[4].Health state utility estimate calculated by applying the disutility generated from the linear mixed-effects model to the overweight health state utility value (0.81) reported by Stephenson et al, 2021[5].
a b Mapping algorithm used in the base-case analysis.cMapping algorithm explored in a model scenario analysis.EQ-5D EuroQol five dimensions health survey, SF-36 36-item short form health survey.

Table S5 : Estimated prevalence of sleep apnoea by model health state based on study by Wall et al, 2012 Model health state Total sample size Number of cases Estimated prevalence b
Publication byWall et al, 2012 [6]reports data for a group defined by a BMI range covering both the obesity I and obesity II model health states.b Calculated as proportion of cases among the total sample. a

Table S6 : Estimated prevalence of non-alcoholic fatty liver disease by model health state based on study by Vusirikala et al, 2020
[7]ublication by Vusirikala et al, 2020[7]reports data for an overall obese group defined by a BMI range covering the obesity I, obesity II, and obesity III model health states.b Calculated as proportion of total number of incident cases across reported phenotype categories among the total sample size across reported phenotype categories for each model health state.

Table S7 : Estimated prevalence of gastro-oesophageal reflux disease by model health state based on study by Jacobson et al, 2012 Number of cases Number of controls Total sample size Estimated prevalence d
[8]lication by Jacobson et al, 2012[8]reports data for a group defined by a BMI range covering both the obesity II and obesity III model health states.
c d Calculated as the proportion of cases among total observations.

Table S8 : Estimated costs of lifestyle management
[11]costs are 2020/21 values.With the exception of clinical psychologist consultations which were incorporated based on feedback from the clinical expert authors, the cost components and their annual frequencies were based on the approach taken in NICE's appraisal of liraglutide for the treatment of obesity[11].

Table S10 : Estimated total comorbidity treatment cost by model health state Model health state Annual cost (£) a
Calculated for each health state by multiplying the annual cost per patient of treating each comorbidity by the estimated prevalence of each comorbidity within each health state. a

Table S11 : Disaggregated base-case results: life years
ESG endoscopic sleeve gastroplasty, LM lifestyle modification, QALY quality-adjusted life year.

Table S13 : Disaggregated base-case results: costs
All costs are 2020/21 values; 3.5% annual discount rate applied to costs and health effects.ESG endoscopic sleeve gastroplasty, LM lifestyle modification.

Probabilistic sensitivity analysis results: incremental cost-effectiveness plane ESG
endoscopic sleeve gastroplasty, ICER incremental cost-effectiveness ratio, LM lifestyle modification, PSA probabilistic sensitivity analysis, QALY quality-adjusted life year.ESG vs LM alone: individual resultsESG vs LM alone: mean PSA result ESG vs LM alone: deterministic ICER