This pilot study will compare hair-based toxicological assessment of cannabis exposure with self-reported use of cannabis over the same historical period amongst a purposive sample of 100 people selected to represent a range of cannabis exposure levels. Self-reported use will be assessed through completion of the assessment schedule currently incorporated in the 15+ ALSPAC Focus clinic (where hair samples are being collected and stored). For convenience, cannabis negative controls (10) will be sought amongst volunteers from the Department of Social Medicine. A further 90 individuals with a range of cannabis use levels from low (monthly or less) to high (daily) will be recruited through advertisements amongst students and users of community drugs projects with whom we have contacts. These sources have been successfully used in the past to recruit to focus groups; participants will be offered a £20 inconvenience fee and will be given assurances of complete confidentiality. Following completion of the questionnaire, an investigator will check participant responses. Participants will then provide a hair sample for toxicology. Assessments of use in the past 3 months provided by toxicology will be compared to those from self-report. Hair samples will be collected via the standard procedures used in ALSPAC.Colour of hair and use of any cosmetic treatments (e.g. perming or bleaching) will be recorded as these can influence toxicology results (though seldom to the point of producing false negatives).

Toxicological assessment will be undertaken at the TrichoTech laboratories in Cardiff. Quantitative assessment of cannabis content of the samples will be undertaken using gas chromatography/mass spectrometry. Cannabis dose exposure as inferred by the amounts and frequencies reported by questionnaire in the same participants will also be calculated and ranked. The two assessments will be compared using weighted Kappa scores to compare categorical assessments of dose/ consumption. Intraclass correlations also will be used to assess association between continuous measures of consumption. Based on a sample size of 100 and consideringthe percentage agreement for a binary outcome, for example "heavy" consumption defined as the top quintile of each distribution, we will be able to detect agreement between the two methods of 60% or more based on the width of the one-sided 95% confidence interval.[i]

High Kappa or correlation scores will be taken as evidence of low levels of reporting bias in the self-reported cannabis measures. Conversely, low agreement between self-report and toxicological assessment will be taken to suggest that the former may have been influenced by social desirability bias.

Results of this pilot study will be used to support funding applications (for example to the NIH and the MRC) for toxicological assessment of the full ALSPAC cohort based on samples collected at the 15+ Focus clinic and subsequently.

[i] Machin D, Campbell M, Fayers P, Pinol A. Sample size tables for clinical studies 2nd Ed. Blackwell, Oxford 1997.