Exploring the perils of cross-national comparisons of drug prevalence: The effect of survey modality

Background: There is significant interest in comparing countries on many different indicators of social problems and policies. Cross-national comparisons of drug prevalence and policies are often hampered by differences in the approach used to reach respondents and the methods used to obtain information in national surveys. The paper explores how much these differences could affect cross-country comparisons. Methods: This study reports prevalence of drug use according to the most recent national household survey and then adjusts estimates as if all national surveys used the same methodology. The analysis focuses on European countries for which the European Monitoring Centre for Drugs and Drug Addiction reports data, the United States, Canada, and Australia. Adjustment factors are based on US data. Findings: Adjusting for modality differences appears likely to modestly affect the rankings of countries by prevalence, but to an extent that could be important for comparisons. For example, general population surveys suggest that the US had some of the highest cannabis and cocaine prevalence rates circa 2012, but this is partially driven by the use of a modality known to produce higher prevalence estimates. This analysis shows that country rankings are partly an artifact of the mode of interview used in national general population surveys. Conclusions: Our preliminary efforts suggest that cross-national prevalence comparisons, policy analyses and, other projects such as estimating the global burden of disease could be improved by adjusting estimates from drug use surveys for differences in modality. Research is needed to create more authoritative adjustment factors.