The 3% - 4% figures is published on DAN's website at:
Estimated DCI Incidence in a Select Group of Recreational Divers
Excerpted:
"Results: Dives from logs, memory, or both had medians estimated from log normal distributions fitted to cumulative frequency plots by eye. Data are given in table. About 75% of responders were male accounting for 82% of dives, the 25% females accounted for 18% of dives. The median for logged dives was similar for males and females but the male median was slightly higher in the other two categories. Fifty-seven divers reported 67 cases of DCI
(4.3% of females, 1.06% of dives; 3.3% of males, 5.3% of dives) (ed.; emphasis added), but only 30 cases were reported as treated. The median number of dives for divers reporting DCI was similar for treated and untreated cases, ~25.2. Whether untreated cases were actually DCI is unknown, but if they aren't, DCI incidence would be only 1.51."
"Conclusion: Total DCI incidence, assuming untreated cases were DCI, is close to Bove's (3.41) for sport divers but above that for military divers (2.65) (Undersea Hyper Med 1998;25:175-178). If the survey population of DAN members is assumed to be representative of the larger population then an average of 1,006 cases of DCI/year, reported to DAN from 1992-1997, at 3.38 DCI/104 dives would represent 2.98 million recreational dives/year. Whether untreated cases represent underreporting or false positives is unknown."
To dismiss or not to dismiss the untreated cases? Hmmm... In their conclusion, the authors indicate that these numbers correlated to another (Bove) study of sport divers. As this thread has shown, even suggesting to the average person that they might be suffering from a mild case of DCI is likely to produce a strong and immediate denial. To systematically dismiss all untreated cases would seem imprudent. At any rate, this information would seem to strongly contradict the .015% number you stated.
One man's "safe" is another man's "scary?" Frankly, I'm not qualified to get into an argument about the relative merits of one study -vs- another. I have just enough of a statistics background to know that all that lies down that road is a headache (sub-clinical DCI?) and wet shoes.
Steven