For temporal trends, country estimates were averaged across 25 countries with full coverage and valid estimates on at least four (including 2015) out of six time points. The countries included are Bulgaria, Croatia, Cyprus, the Czech Republic, Denmark, Estonia, the Faroes, Finland, France, Greece, Hungary, Iceland, Ireland, Italy, Lithuania, Malta, the Netherlands, Norway, Poland, Portugal, Romania, Slovakia, Slovenia, Sweden and Ukraine. The average across the 25 country means was calculated using a weight of 1, and data for each survey year were summed and divided by the number of countries with valid data for that particular year.

Data across all six time points were complete in 20 countries with data on five countries missing in 1995 (Bulgaria, France, Greece, the Netherlands and Romania), and data on one country missing in 1999 (the Netherlands; data collected but not considered comparable) and 2007 (Denmark; data collected but not considered comparable). Trends across the 25 countries are shown for a number of selected indicators by gender. Country-specific trends are shown for all countries with at least two valid data points over the period 1995-2015. Country-specific prevalence estimates before the 2007 ESPAD survey are only available on aggregate level, preventing statistical testing of temporal changes. Instead, a standardised procedure is used where a difference of less than ± 3 percentage points is not considered as ‘real difference’. Trends are illustrated graphically, with decreases of 3 or more percentage points between successive surveys indicated in green, increases of 3 or more percentage points in red and unchanged situations in yellow (less than ± 3 percentage points).

Comparability of variables

After the 2003 survey, a working group was set up to improve and revise some of the questions that had caused problems in the previous surveys. Modified questions were tested on differences in outcome using a split-half design in eight countries. In general, most of the revised questions were found to be comparable with the earlier versions (Hibell and Bjarnason, 2008).

Availability of substances

In the surveys until 2003, perceived availability of substances was asked in one single question. Since 2007, the questionnaire has contained separate questions for each substance. A questionnaire test in eight countries showed some differences between the two versions.

Alcohol use

In the surveys until 2003, the question on heavy episodic drinking read ‘How many times (if any) have you had five or more drinks in a row? A “drink” is a glass of wine (approximately 15 centilitres), a bottle or can of beer (approximately 50 centilitres), a shot glass of spirits (approximately 5 centilitres) or a mixed drink.’ Cider or alcopops were not included. Since 2007, the definition has read: ‘How many times (if any) have you had five or more drinks on one occasion? A “drink” is a glass/bottle/ can of beer (approximately 50 centilitres), a glass/bottle/ can of cider (approximately 50 centilitres), two glasses/ bottles of alcopops (approximately 50 centilitres), a glass of wine (approximately 15 centilitres), a glass of spirits (approximately 5 centilitres) or a mixed drink.’ The questionnaire test revealed no significant differences between the two versions.

Illicit drugs other than cannabis

The questionnaire collects data on the use of illicit drugs other than cannabis, including amphetamine, cocaine, crack, ecstasy, LSD or other hallucinogens, heroin, GHB (since 2007) and methamphetamine (since 2015). In 2015, crack was not included in Denmark, Estonia, Finland and Sweden.

Inhalant use

In the earlier rounds of the survey, the question on inhalant use read ‘Did you try inhalants (glue, etc.) to get high?’. In 2007, the question was rephrased to refer to ‘the use of inhalants to get high’. The questionnaire test found no significant differences between the old and new versions. Since 2011, countries have been instructed to add nationally relevant examples in the questionnaire.