matrix of likely sample size based on these values ranged from 44 to 193 per group. Based on this DPP, a sample of 120 per group was selected as the sample size that would deliver clinically meaningful results. ConClusions: A DPP is useful in late phase research to define appropriate sample size where no data exist. It is important to note that DPP methods do not require significance testing, but the benefit is no need for a correction for multiple comparisons at the time of the final analysis.

A503(vPDT) fell from 93% to 68%. In the 'without ranibizumab' scenario, only vPDT was administered. Costs of treatment, administration, monitoring, bilateral disease and management of recurrences were included. Results: An estimated 2045 patients were eligible for treatment at year 1 and 2119 at year 5. In the 'with ranibizumab' scenario, 143 patients received ranibizumab at year 1, increasing to 678 at year 5; 1902 patients received vPDT at year 1 and 1441 at year 5. 'With ranibizumab' annual costs were higher in years 1-2 than 'without ranibizumab' costs. During years 3, 4 and 5, cost savings occurred with ranibizumab (£3867, £121 584 and £232 467, respectively), owing to lower total costs of treatment and monitoring than with vPDT. The total 5-year saving 'with ranibizumab' over 'without ranibizumab' was £227 086. ConClusions: Treating visual impairment due to CNV secondary to PM with ranibizumab rather than vPDT is estimated to provide significant cost savings in England and Wales over 5 years.