Abstract
The interview is often used in case-control-studies to get information
about a possible hazardous exposition. Especially if the recall period is very
long a recall bias cannot be excluded.
Validation studies about the interview data regarding the use of oral
contraceptives done so far seem to show that the quality of recall is sufficient
enough for a risk calculation in a case control study. No such validation
study was done in Germany so far.
This validation study was done to get an estimate about the quality of
data regarding the use of oral contraceptives that came from a
multicentre-liver-tumour study (MILTS), collected by standardised personal interview in
Germany.
Additionally the validation study investigated whether a mass media
campaign in Germany about a potential liver toxicity of cyproterone acetate had an
influence on the recall of these medications. Validation of the interview
data regarding the use of oral contraceptives was done using the data of the
medical records of the oral contraceptive prescribing physicians.
The results seem to show that women in Germany recall the lifelong use of
oral contraceptives sufficiently to estimate the risk for a rare event like
hepatocellular carcinoma in a case-control-study. The agreement on ever-use
and on duration of use for the validated time parts was good. There was no
sign of differential recall between cases and control. The agreement regarding
special brands of oral contraceptives was less good especially for the proxy
interviews.
Although the number of ever-user of cyproterone acetate was low there was
no hint that a biased recall of these medications led to an underestimate of
the risk of liver cancer in the multicentre liver tumour study. But it
cannot be excluded that the mass media campaign led to a higher response rate in
these women. This would have led to a higher estimate of the risk of liver
cancer in the case-control-study.
Considering the situation in Germany regarding the medical treatment, the
documentation and storing of medical records and data security these kind of
validation studies about the lifelong use of medications with the medical
record as a standard should only be carried out for very important questions.
Since the German situation is not suited to perform a routine validation, the
validation will always only be possible for certain time periods with many
gaps and will be associated with high costs. | 
