Lead-time bias
Lead-time bias refers to a spurious increase in longevity
associated with screening. For example, assume that
mammography screening leads to cancer detection 2 years
earlier than would have ordinarily occurred, yet the
screening does not prolong life. On average, women with
breast cancer detected through screening live 2 years
longer than those with cancers diagnosed through
traditional means. This gain in longevity is apparent and
not real: this hypothetical screening allows women to live
2 years longer with the knowledge that they have cancer,
but does not prolong survival, an example of zero-time
shift.2
Length bias
Length bias is more subtle than lead-time bias: the
longevity association is real, but indirect. Assume that
community-based mammography screening is done at
10-year intervals. Women whose breast cancers were
detected through screening live 5 years longer on average
from cancer initiation to death than those whose cancers
were detected through usual means. That screening is
associated with longer survival implies clear benefit.
However, in this hypothetical example, this benefit
indicates the inherent variability in cancer growth rates
and not a benefit of screening. Women with indolent,
slow-growing cancers are more likely to live long enough
to be identified in decennial screening. Conversely, those
with rapidly progressing tumours are less likely to survive
until screening.
The only way to avoid these pervasive biases is to do
randomised controlled trials and then to assess agespecific
mortality rates for those screened versus those not
screened.10 Moreover, the trials must be done well. The
quality of published trials of mammography screening has
raised serious questions about the utility of this massive
and hugely expensive enterprise.22–24
Conclusion
Screening can promote or impair health, depending on its
application. Unlike a diagnostic test, a screening test is
done in apparently healthy people, which raises unique
ethical concerns. Sensitivity and specificity tend to be
inversely related, and choice of the cut-off point for
abnormal should indicate the implications of incorrect
results. Even very good tests have poor predictive value
positive when applied to low-prevalence populations.
Lead-time and length bias exaggerate the apparent benefit
of screening programmes, underscoring the need for
rigorous assessment in randomised controlled trials before
use of screening programmes.
Acknowledgments
We thank Willard Cates and David L Sackett for their helpful comments
on an earlier version of this report. Much of this material stems from our
15 years of teaching the Berlex Foundation Faculty Development Course.