The recent availability of World Wide Web-based directories has opened up a
new approach for tracing subjects in epidemiologic studies. The completene
ss of two World Wide Web-based directories (Canada411 and InfoSpace Canada)
for subject tracing was evaluated by using a randomized crossover design f
or 346 adults randomly selected from respondents in an ongoing cohort study
. About half (56.4%) of the subjects were successfully located by using eit
her Canada411 or InfoSpace. Of the 43.6% of the subjects who could not be l
ocated using either directory, the majority (73.5%) were female. Overall, t
here was no clear advantage of one directory over the other. Although Canad
a411 could find significantly more subjects than InfoSpace, the number of p
otential matches returned by Canada411 was also higher, which meant that a
longer list of potential matches had to be examined before a true match cou
ld be found. One strategy to minimize the number of potential matches per t
rue match is to first search by InfoSpace with the last name and first name
, then by Canada411 with the last name and first name, and finally by InfoS
pace with the last name and first initial. Internet-based searches represen
t a potentially useful approach to tracing subjects in epidemiologic studie
s.