A661for 40% of decisions, were compared with those of three major HTA agencies: CADTH, NICE and PBAC. HTA reports and meeting transcripts were analysed and categorised by date, therapy area, decision, rationale, and pricing decision. Resubmissions or those not assessed by the western HTA agencies were excluded. Results: A total of 65 NHI reports were identified. Of these 26 reported decisions on oncology or cardiovascular drugs, 12 were excluded (3 resubmissions, 9 not reviewed by the other agencies). Prior to 2GNHI, 4 out of 5 decisions were positive, or 80% approval rate, while after, only 4 out of 9 were positive, a 44% approval rate. Prior to 2GNHI, all NHI reimbursement decisions identified (6) matched CADTH, NICE, and PBAC. After 2GNHI, only 6/9 or 66% matched. Clinical effectiveness and budget impact were most cited in reimbursement rejections. For example Zytiga, NHI appreciated the cost-effectiveness but stated budget impact was too high, issuing a negative recommendation, contrary to the other agencies. Interestingly, a 'local' product was recommended for limited reimbursement even though budget impact was high. ConClusions: Since implementation of Taiwan's NHI reforms in January 2013, cardiovascular and oncology drug approvals dropped by 36% and agreements with western agencies down 34%, placing an emphasis on budget impact. However, this analysis was constrained by its small sample size, and limited therapy areas.objeCtives: Phase III, randomised controlled trials remain the gold standard for health technology assessment (HTA) submissions. Data sets may be supplemented with other sources (e.g. Phase II trials, observational studies, mixed treatment comparisons). However, the influence of expanded data sets on HTA appraisals is unclear. Methods: We reviewed recent National Institute for Health and Care Excellence (NICE; England and Wales) oncology drug submissions to determine the frequency and type of expanded data sets. We then evaluated the influence of expanded data on agency decisions. A similar review of submissions to the Australian Pharmaceutical Benefits Advisory Committee (PBAC) was performed. Results: There were 30 relevant appraisals on the NICE website covering a range of cancer types. Of these, 14/30 made use of expanded data sets featuring Phase II trials, observational studies, meta-analyses and/or mixed treatment comparisons among other sources. Reasons for using expanded data sets included: agency concerns over Phase III studies, lack of long-term or head-to-head data and limited Phase III data. Where additional data were included, around one third (5/14 cases, 35.7%; 5/30 [16.7%] overall) appeared to have directly influenced the final decision. Overall, positive appraisals were less frequent for submissions that featured expanded data sets compared with submissions featuring Phase III data only (2/14 [14.3%] versus 9/16 [56.3%]). Comparable to NICE, 2/10 (20%) of PBAC submissions were influence by expanded data sets, although cost-effectiveness data were crucial for PBAC approval ove...