The Committee also considered the estimates of cost included in the manufacturer's economic analysis, particularly those relating to the number of cycles of treatment and the inclusion of non-treatment costs (including BSC). The Committee concluded that the estimate of clinical effectiveness and of the number of treatment cycles should be based on the same source – that is, the registration trial in which a mean of 4.4 cycles was used. It noted that even if, in clinical practice, the number of cycles used is lower than this, the clinical effectiveness of shorter regimens (for example, the 3 cycles of treatment which were included in the manufacturer's analysis) was unknown. The Committee also considered the appropriateness of the estimates of patient average BSA. The Committee acknowledged that BSA would vary between patients and concluded that the mean BSA could be lower than the ERG estimate, particularly in patients with relapsed NSCLC. The Committee considered the manufacturer's estimate to be appropriate, but concluded that this factor would not substantially change the ICER. The Committee noted that if the ERG's revised estimates of costs were included, the ICER for pemetrexed compared with docetaxel would be greater than £1 million per additional QALY gained.