BackgroundTwo approaches to the procurement of recombinant Factor VIII products are used by health systems: (A) the most common approach where acquisition tenders are not carried out; (B) the approach tested in the UK in which procurement is based on tenders. The respective cost-effectiveness is not known.ObjectiveTo estimate the incremental cost-effectiveness ratio (ICER) for the comparison A vs B.MethodsThe analysis evaluated: (i) Factor VIII cost with/without tenders; (ii) inhibitor development caused by switching between products; (iii) clinical and economic consequences of inhibitors. Information on these items was obtained from a literature search. Because of the scarce evidence available on some items, our analysis considered the ‘most favourable’ scenario—that is, some extreme though reasonable assumptions were adopted that were intentionally biased towards improving the ICER of the no-tender option.Results and discussionWe estimated an ICER for A vs B of £486 409 (€657 139; £1=€1.351) per quality-adjusted life year (QALY). Since pharmacoeconomic thresholds are ∼£30 000 per QALY, our results indicate that the cost-effectiveness of acquisition strategies that avoid tenders is prohibitive. Because of the simplified nature of our analysis, this estimate is preliminary.ConclusionsThe ‘true’ ICER of A vs B remains unknown, but its value is likely to be even worse than the unfavourable ICER of £486 409 (€657 139) per QALY.
Linear and quadratic discriminant analysis are considered in the small-sample, high-dimensional setting. Alternatives to the usual maximum likelihood (plug-in) estimates for the covariance matrices are proposed. These alternatives are characterized by two parameters, the values of which are customized to individual situations by jointly minimizing a sample-based estimate of future misclassification risk. Computationally fast implementations are presented, and the efficacy of the approach is examined through simulation studies and application to data. These studies indicate that in many circumstances dramatic gains in classification accuracy can be achieved.