ABSTRACT
For studying trends in blood biochemistry analytes of an individual or a group of individuals, the outcome may be influenced by analytical changes that may have occurred during the study. An observed trend may well represent a drift in analytical performance instead of a truly biological finding. We developed a model that allows for retrospective correction of analytical changes with time. This model is based on the concept of adjustment of an individual's longitudinal blood biochemistry data by comparing the long-term results of the laboratory with those of other laboratories in an external quality-control survey program. Factors responsible for the analytical bias of our laboratory were identified by multiple regression analysis. The resulting procedure for assessing analytical bias and variability was applied to study in two mutually exclusive cohorts of employees of the Shell petrochemical complex in Rotterdam (a) the true nature of the changes (analytical or biological) in gamma-glutamyltransferase (GGT) and (b) the effect of age on GGT. The first cohort consisted of employees who attended a periodic health assessment in 1984 and in 1989; the second, employees who attended periodic health assessments in 1985 and in 1988. Thus we studied 3- and 5-year changes of GGT corrected for analytical bias. Whereas standard cross-sectional results apparently showed an increase of GGT up to age 50 years, the longitudinal findings corrected for analytical changes, as indicated above, do not support these cross-sectional results.
