Table 2. Projection of Prostate Cancer Mortality Reduction in a Hypothetical Screening Trial Ignoring and Preserving Subtypes According to their Distribution Among Late-Stage Diagnoses.

Box 1: Inputs to Mortality Reduction Projection
	Low Grade	High Grade	Total
(A) Years of follow-up			10
(B) Incidence Hazard			0.005
(C) Distribution of Cases
Early Stage	66%	13%	80%
Late Stage	8%	13%	20%
Total	74%	26%	100%
(D) Mortality Hazard
Early Stage	0.017	0.087	0.026
Late Stage	0.161	0.258	0.209
Total	0.026	0.138	0.045

Box 2: Mortality Reduction Projection
	Low Grade	High Grade	Total
(E) Stage Shift Multiplier	0.399	0.292	0.437
(F) Stage Shift	11%	19%	30%
Mortality Reduction
(G) Projected Ignoring Grade			13%
(H) Projected Preserving Grade	5%	5%	10%
(I) Absolute Difference			+3%
(J) Relative Difference			+31%

• (A)
  Years of follow-up for hypothetical trial
• (B)
  Incidence hazard of prostate cancer estimated from SEER (9 registries, Nov 2004 submission) for men ages 50-59 years in 1980-1984
• (C)
  Distribution of cases from SEER for men diagnosed ages 50-59 years in 1980-1984
• (D)
  Hazard of mortality following diagnosis, derived from SEER 10-year survival rates, assuming a constant hazard of mortality.
• (E)
  Stage-shift multipliers calculated as in Owens et al. [5], see formulas G and H in that text. The inputs to these values are the length of follow-up (A), overall incidence rate of the disease (B), the rates at which the disease is diagnosed in each stage (C), and stage-specific survival rates (D). The calculation is performed with all grades combined, and then separately by grade using the corresponding inputs. This value is used below to convert a percent reduction in late-stage disease (F) to a percent reduction in disease-specific mortality (G, H).
• (F)
  Percent reduction in late-stage cases. Hypothetical value of 30% in screening arm, allocated between grades proportional to late-stage case mix (C).
• (G)
  = (E) x (F) in total
• (H)
  = (E) x (F) for each grade separately and summed across grades
• (I) and (J)
  Absolute and relative differences of projection in total compared to projection by grades