N-acetylglucosamine-6-sulfate sulfatase deficiency reconsidered.

design (3). We can imagine that some statistical procedure more powerful than standard Mantel-Haenszel and Miettinen tests might narrow the confidence intervals a bit, but it would not alter these point estimates by much. Adjustment of the RR's for tobacco use provides an adequate test for confounding. With subjects categorized either as non-, ex-, or current cigarette smokers or as pipe or cigar smokers, the adjusted RR's were 0.97 for males and 0.76 for females, practically unchanged from their original values. Thus, tobacco use was not a confounder. We appreciate Bross's concern that we not overlook potential artifacts or alternative interpretations of our data, but having reexamined them we see no reason to change any of our conclusions. ERNST L. WYNDER American Health Foundation, 320 East 43 Street, New York 10017 STEVEN D. STELLMAN American Cancer Society, Inc., 777 Third Avenue, New York 10017