I bet my 2.5 grams of red blood cells that more people have died from transfused diseases and adverse reactions than from refusal, and as the Australian government is realizing, its not the best option, especially now.
I don't know how we could possibly resolve such a bet, (And it would be macabre anyway) but a comparison of applicable studies strongly contradicts this idea.
From a JW perspective, the most favorable study was done by Kitchens. He compiled 1404 cases involving Jehovah's Witnesses and collated them into 16 surgical catagories Short-term mortality directly attributable to refusing blood ranged from 0 to 8.33%, depending on the type of surgery. Overall, increased mortality from refusing blood was 0.5% to 1.5% (Kitchens, C. Are Transfusions Overrated? Am J Med 1993;94:117-119)
On the flip-side of the coin, the risk of accepting blood was covered in a study by Sazama. Here, the short-term mortality rate, from accepting blood was 1 to 1.2 per 100,000 patients who received blood transfusions. In other words, receiving blood transfusion increased the mortality by 0.001 to 0.0012%, whereas refusing blood transfusion increased the mortality by 0.5% to 1.5%. (Sazama, K.: Reports of 355 Transfusion-Related Deaths: 1976 - 1985 Transfusion 1990;30:583-90)
The risk of blood transfusion was again extensively reviewed in the "Medical Progress" review in the New England Journal of Medicine in 1999. In this progress review, the overall number of deaths from blood transfusion is estimated at 23 and 44 deaths per million units of blood. These numbers include every complications from blood transfusion, not just short-term mortality as in Sazama's report. The short-term mortality which corresponds to Kitchens' estimate should be smaller than these numbers. Thus, a typical blood transfusion of two units carries the risk of 46 to 88 overall deaths per million patients, or 0.0046 to 0.0084%. If a larger amount of blood is transfused, this risk increases further, to 0.01 to 0.03%. (Goodnough LT, Brecher ME, Kanter MH, et al.:Transfusion Medicine. First of two parts--blood transfusion N Engl J Med 1999;340:438-447)
Based on this data, the risk of refusing blood is greater than the risk of accepting blood by an order of magnitude. --Roughly 100 times greater.