The 40 percent of data missing from this graph represented the subset of patients in the study who had the largest decreases in bone turnover (60 to 80 percent). If P&G was right that Actonel and Fosamax were equally effective, then this group of patients should have had roughly the same number of bone fractures as patients with far lower reductions in bone turnover (the 30-to-40-percent levels P&G cited). But, according to Blumsohn, the data revealed just the opposite. The patients with the highest reductions in bone turnover—who were excluded from the graph—experienced fewer fractures than the average patient taking Actonel. Most had no fractures at all.