CASE DEFINITION IN CASE-CONTROL STUDIES OF THE EFFICACY
OF SCREENING
ALAN S. MORRISON

Consider a case-control study that focuses
on mortality. Eligible cases would
be deaths from the disease screened for
irrespective of the means of initial diagnosis.
This group would include deaths
from disease diagnosed as a result of
screening, as well as deaths from disease
that was first diagnosed following the occurrence
of symptoms. Eligible controls
would include all living members of the
source population (5)—people who never
had the disease, as well as people who had
had the disease, whether or not screendetected.

It may not be possible to obtain satisfactory
screening histories from records,
and it may be necessary to interview living
cases, and controls, to determine
screening histories. In this situation, potential
cases could be defined as people
who develop a manifestation of the disease
(other than death) that is sufficiently advanced
to ensure diagnosis even without
screening, and thus be past the lead-time
interval. For example, cases could be defined
as people who develop symptomatic
metastases of cancer. Such people would
include those in whom screening detected
early cancer that later recurred and
spread as well as people with advanced
cancer whose disease was first diagnosed
following the occurrence of symptoms.
Members of the source population without
the defining characteristic of a case—i.e.,
advanced disease—would be eligible for
the control series. Those eligible would
include people who had had early disease,
whether or not screen-detected, and
whose disease had been treated and had
not recurred, as well as people who had
never had the disease, whether or not
they had been screened. The proportions
screened in the case and control series yield

an estimate of the ratio of the rate
of development of advanced cancer
in screened compared to unscreened people
(5).