Protecting Privacy of Shared Epidemiologic Data without Compromising Analysis Potential

Table 1

Cases in the distributed data matching to ≤3 doses in the source database.

Comparison to	Number of matches in source database (identification risk)			Overall identification risk (1)
	3 (0.33)	2 (0.5)	1 (1.0)
	(a) Using original (nonmasked) doses
All organ doses	2	9	18	0.39
Stomach doses only	3	5	29	0.52
AHS stomach doses	1	3	35	0.58
	(b) Using doses rounded to 3 significant digits
All organ doses	0 (1)^a	0 (5)	0 (6)	0 (0.14)
Stomach doses only	0	0	0 (2)	0 (0.03)
AHS stomach doses	0	0	0	0 (0)

^aNumbers in parentheses are the numbers of values that exactly matched some entry in the source database, but in no instance was the matching individual in the source database the subject with the three-digit rounded dose.