Bg. Armstrong, EFFECT OF MEASUREMENT ERROR ON EPIDEMIOLOGIC STUDIES OF ENVIRONMENTALAND OCCUPATIONAL EXPOSURES, Occupational and environmental medicine, 55(10), 1998, pp. 651-656
Random error (misclassification) in exposure measurements usually bias
es a relative risk, regression coefficient, or other effect measure to
wards the null value (no association). The most important exception is
Berkson type error, which causes little or no bias. Berkson type erro
r arises, in particular, due to use of group average exposure in place
of individual values. Random error in exposure measurements, Berkson
or otherwise, reduces the power of a study, making it more likely that
real associations are not detected. Random error in confounding varia
bles compromises the control of their effect, leaving residual confoun
ding. Random error in a variable that modifies the effect of exposure
on health-for example, an indicator of susceptibility-tends to diminis
h the observed modification of effect, but error in the exposure can c
reate a spurious appearance of modification. Methods are available to
correct for bias (but not generally power loss) due to measurement err
or, if information on the magnitude and type of error is available. Th
ese methods can be complicated to use, however, and should be used cau
tiously as ''correction'' can magnify confounding if it is present.