Multi-tier drugs assessment in a decentralised health care system. The Italian case-study

Objective To investigate the organisation and decision-making processes of regional and local therapeutic committees in Italy, as a case-study of decentralised health care systems. Methods A structured questionnaire was designed, validated, and self-administered to respondents. Committee members, prioritisation, assessment process and criteria, and transparency of committees were investigated. Results The respondents represent 100% of the 17 regional committees out of 21 regions (in 4 regions there is not any regional formulary), 88% of the 16 hospital networks and 42% of the 183 public hospitals. The assessment process appears fragmented and may take a long time: drugs inclusion into hospital formularies requires two steps in most regions (regional and local assessment). Most of the therapeutic committees are closed to industry and patients associations involvement. Prioritisation in the assessment is mostly driven by disease severity, clinical evidence, and the absence of therapeutic alternatives. Only 13 out of the 17 regional committees have a public application form for drugs inclusion into regional formulary. Regional and local committees (i) often re-assess the clinical evidence already evaluated at central level and (ii) mostly rely on comparative drug unit prices per DDD and drug budget impact. The level of transparency is quite low. Conclusions The Italian case-study provides useful insights into an appropriate management of multi-tier drugs assessment, which is particularly complex in decentralised health care systems, but exists also in centralised systems where drugs are assessed by local therapeutic committees. A clear definition of regulatory competences at different levels, a higher collaboration between central, regional and local actors, and increased transparency are necessary to pursue consistency between central policies on price and reimbursement and budget accountability at the regional and local levels.