Article type
Year
Abstract
Background: Systematic reviews of the effectiveness of health care interventions are the 'gold standard' for evidence of effectiveness. Whether an intervention should be implemented in practice depends also on its costs. Whether an intervention is 'cost-effective' depends on the statistical uncertainty around costs and effects and the maximum willingness to pay. Modelling with a monte-carlo analysis, based on the mean and variance of the pooled effect estimate may enable the uncertainty in the trial data to be reflected in estimates of cost-effectiveness. This study aimed to determine the likely cost-effectiveness of H.pylori eradication therapy in patients with non ulcer dyspepsia.
Methods: Meta-analysis of RCTs. Eligibility criteria; adequate definition of dyspepsia; exclusion of peptic ulcer or oesophagitis by endoscopy; effective H.pylori eradication regime; ITT analysis of effect on dyspeptic symptoms. Markov model of cost-effectiveness of intervention v. antacid therapy alone over 1 year (Data 3.5, Treeage software). Second order Monte-carlo analysis on relative risk, represented as a lognormal distribution with mean and variance from the meta-analysis. Results, and sensitivity analysis on costs, presented as a cost-effectiveness acceptability curve using Excel.
Results: Nine trials with a total of 2,541 patients were pooled. The relative risk of dyspepsia after H.pylori eradication was 0.91 (95% CI 0.86 - 0.96). The modelling indicated that there would be an 80% chance that the intervention would be cost-effective at a maximum willingness to pay of 75 per month free of dyspepsia. Sensitivity analysis on the cost of eradication therapy indicated that if the cheapest eradication was used the maximum willingness to pay could fall to 25, at the same level of uncertainty.
Conclusions: This method allows the cost-effectiveness of an intervention to be estimated with the same statistical rigour as the underlying effect data, yet extends the usefulness of systematic reviews for those responsible for implementing their findings.
Methods: Meta-analysis of RCTs. Eligibility criteria; adequate definition of dyspepsia; exclusion of peptic ulcer or oesophagitis by endoscopy; effective H.pylori eradication regime; ITT analysis of effect on dyspeptic symptoms. Markov model of cost-effectiveness of intervention v. antacid therapy alone over 1 year (Data 3.5, Treeage software). Second order Monte-carlo analysis on relative risk, represented as a lognormal distribution with mean and variance from the meta-analysis. Results, and sensitivity analysis on costs, presented as a cost-effectiveness acceptability curve using Excel.
Results: Nine trials with a total of 2,541 patients were pooled. The relative risk of dyspepsia after H.pylori eradication was 0.91 (95% CI 0.86 - 0.96). The modelling indicated that there would be an 80% chance that the intervention would be cost-effective at a maximum willingness to pay of 75 per month free of dyspepsia. Sensitivity analysis on the cost of eradication therapy indicated that if the cheapest eradication was used the maximum willingness to pay could fall to 25, at the same level of uncertainty.
Conclusions: This method allows the cost-effectiveness of an intervention to be estimated with the same statistical rigour as the underlying effect data, yet extends the usefulness of systematic reviews for those responsible for implementing their findings.