RT Journal
A1 Palen TE
T1 PErsonal health records and medical care use—reply
JF JAMA
JO JAMA
YR 2013
FD February 27
VO 309
IS 8
SP 767
OP 768
DO 10.1001/jama.2013.384
UL http://dx.doi.org/10.1001/jama.2013.384
AB 
I also agree that volunteer bias is a potential problem not only for retrospective studies such as ours, but also for all observational research studies. We attempted to control for volunteer bias in our study by having each cohort serve as their own control and by performing both intracohort and intercohort analyses. Volunteer bias is controlled within a trial by randomization, but it still may occur when patients, clinicians, or sites decide whether to participate.