Background: The utility of using biomarkers of smoking in epidemiological studies depends not only on the validity and precision of the laboratory procedure but often on the long-term stability of the analytes of interest in stored biological samples.
Methods: We retrieved urine samples collected in 1976-1977 from women included in a cohort study in Utrecht and for whom information on smoking status was available. Creatinine and thiocyanate were measured in 1976-1977 on fresh samples. Cotinine and creatinine were analysed in 1988 on urine stored at -20 degrees C.
Results: Measurements of creatinine more than 10 years apart showed a correlation of 0.95 and equal means. Cotinine measurements made in 1988 allowed a clear separation of smokers and non-smokers (sensitivity 92%, specificity 100%), suggesting that concentrations retained their discriminant value even after 10 years of storage.
Conclusion: These results emphasize the possibilities offered by long-term storage, under proper conditions, of biological samples for subsequent determination of analytes which may emerge as the study progresses.