[Reliability and validity of self-reported drug use among secondary school students]

To assess the reliability and validity of self-reported use among secondary school students.

Validity was assessed in a representative sample of nearly 1,300 students by analyzing: a) the proportion of questions on drug use left unanswered compared with that for other questions; b) the proportion of inconsistencies between related questions; c) the proportion of questions wrongly completed; d) admision of ficticious drug use; e) the relationship between self-reported drug use and that of friends', and f) willingness to admit cannabis and ecstasy use. Reliability was analyzed using the kappa index, the proportion of specific agreement, and the intra-class correlation coefficient in a test-retest procedure in a randomized subsample of 349 students.

The response rate to questions on drug use was high and was similar to that for more neutral questions. Only 0,3% of the secondary school students reported having used a fictitious drug. Except in the case of heroin, individual drug use was directly related to friends' perception of consumption. A very low proportion of students would not be willing to admit to use of cannabis (2%) or ecstasy (3,4%). Questions referring to drug use at some time during students' lives showed greater reliability than those referring to more recent drug use (last 12 months). The kappa indexes for drug consumption at some time during students lives ranged from 0.65 to 0.87, except for ecstasy and LSD (0.51 and 0.52 respectively). The age of first drug consumption was highly reliable (intra-class correlation coefficients ranged from 0.71 to 1).

Except for ectasy, amphetamines and LSD, indexes of reliability and validity were generally good and similar to those obtained in other studies. These findings support the idea that information about self-reported drug use obtained through a questionnaire is reliable and valid, although the absolute prevalences of the use of some drugs should be interpreted with caution. The indicator of consumption at some time during students' lives is especially useful in studies monitoring drug use among this population.