Metodi statistici per la valutazione di costo-efficacia in HTA

Health technology assessment (HTA) is a potential multidisciplinary tool for decision-making in healthcare policy. Scientific evidences for HTA processes usually come from randomized clinical trials (RCTs), which are the gold standard, but observational studies can be used to overcome study design limitations of RCTs. Nowadays, the use of healthcare administrative databases (HADs) in HTA is not widespread, despite their apparent utility. The objective of this work was to use HADs to conduct a cost-effectiveness evaluation of a pharmacological therapy under real world adherence to treatment. We analyzed the case of statin therapy in the secondary prevention of acute myocardial infarction (AMI). Data were derived from HADs of the healthcare authority of Lombardy, a region in Northern Italy. Patients hospitalized for a first event of AMI in the year 2003 were followed until the 31st December 2008, collecting data on healthcare services and vital status. We used regression methods to estimate the incremental net benefit of treatment or of adherence levels to treatment. HADs provide a rich source of information but the analysis must account for the correlation between repeated measures of the same subject, possible selection bias and endogeneity of the data. In this situation classical regression methods are not appropriate, so we applied marginal structural models (MSMs) to the health economics framework. The cost-effectiveness analysis of statin treatment in the secondary prevention of AMI confirms that statins are cost-effective while the analysis of adherence to treatment needs further assessments. The implementation of MSMs could be a future tool in economic healthcare evaluations, because MSMs make possible to account for time-varying confounding and selection bias typical of longitudinal observational studies. In this way the use of data from HADs or from observational studies could become more reliable.